ARTIFICIAL INTELLIGENCE RESEARCH IN THE USSR

• Document Type: FOIA
• Collection: FOIA Collection
• Document Number (FOIA) /ESDN (CREST): 06731721
• Release Decision: RIPPUB
• Original Classification: U
• Document Page Count: 67
• Document Creation Date: December 28, 2022
• Document Release Date: April 5, 2018
• Case Number: F-2017-00295
• Publication Date: July 8, 1964

Approved for Release: 2018/03/28 C06731721
Copy No;
.111.0111....13.EIZIMWMIreff�
Distribution List for OSE- SRAy-37
Title
r
Classification
Distributed c-eif
Analyst
Production Specialist
No. of Copies
1
1
6
Si,MD
api6 ea,,,,4A.y0A)44.66,t4, oDOP
ecipierit
PPB/IPS
(b)(3)
(b)(6)
(b)(3)
(b)(6)
L.s0(Originating Div )
NFD
BMW
DSD
GSD
LSD
History File
(b)
(3)
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
ARTIFICIAL-INTELLIGENCE RESEARCH
IN THE USSR
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
PREFACE
Artificial-intelligence is a popular term that was coined
in the United States during the 1950's to categorize research
studies aimed at simulation of intelligent or "thinking" be-
havior. This type of research seeks to analyze the factors in-
volved in the making, by humans, of specific types of decisions,
to specify and define these factors as decision procedures or
mathematical algorithms, and ultimately to fabricate hard-
ware which can concretely model decision procedures and
thereby assist human decision makers in making increasingly
complex decisions. Although current digital computers can as-
sist in the solving of some complex decision problems, artificial-
intelligence research has already discovered decision routines
which, while they can be modeled on a digital computer for
demonstration purposes, are more efficiently solved with other
types of hardware. In the future this will increasingly involve
some sort of analog or combined analog-digital equipment, pos-
sibly a replica of the structure of the human brain, more likely
a model abstracting essential elements of brain function on
principles not yet uncovered.
Artificial-intelligence research is necessarily interdiscipli-
nary in nature, involving such traditional areas as biology, phys-
iology, psychology and electrical and systems engineering, with
a strong under-laying of mathematics. As used in US literature,
the term "artificial intelligence" may appear to be a rather
flexible "tent," encompassing more or less whatever one desires
to place within it Nevertheless, artificial -intelligence is a "far, .
out" field of scientific endeavor, the surface of which has been
merely scratched. Its potential for future accomplishments
can be only dimly seen and not evaluated at present. If it
succeeds in significantly optimizing decision making in such
complex areas as the economy or national strategic planning,
it will obviously make a strong contribution to a relatively mon-
olithic system, such as the Soviet one.
In the USSR, research on approaches to the fabrication of
problem-solving or decision-making machines (that is, artificial-
intelligence research) is conducted under the general category
of cybernetics. This report covers Soviet research on major
problems in this field, including pattern recognition by ma-
chines, machine learning, planning and induction in the prob-
lem-solving machine, and brain modeling. It does not cover
conventional digital computer solution of problems or on-line
computer control of processes.
� ..The-material in this report is based-chiefly on-information" -
from the Soviet open literature available as of 1 May 1964.
Additional information obtained through 1 July 1964 does not
materially affect the conclusions.
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721_
ARTIFICIAL-INTELLIGENCE RESEARCH
IN THE USSR
PROBLEM
To assess artificial-intelligence research in the USSR
CONCLUSIONS
1. In the Soviet Union, substantial govern-
mental and Party support and encouragement
are being given to extensive studies on the
theory of artificial-intelligence. Soviet sci-
entists of high caliber are conducting arti-
ficial-intelligVide "Itirdieg� and- are exposing
young students to the state of the art in
this interdisciplinary field, laying a sound
foundation for further advances.
2. The importance which the Soviet regime
attaches to artiflcial-intelligence research is
attested by the unusual freedom of open dis-
cussion allowed scientists working in the field,
as reflected in the published literature.
3. Soviet research in the theoretical and
hardware aspects of artificial-intelligence is
now about as advanced as -US-work and can �
be expected to continue at a rate equal to or
greater than that observed in the West. Sig-
nificant theoretical achievements within the
next 5 years are highly probable. When
theories are converted into designs, Soviet en-
gineers probably will be able to produce the
equipment.
SUMMARY
After a relatively late start, Soviet research
on artificial-intelligence related to the ulti-
mate development of decision-making ma-
chines now is about on a par with similar US
work.---The apparently -greater rate of Soviet
progress compared with that of the West is
attributable to the magnitude of official rec-
ognition and support of artificial-intelligence
research in the USSR. Soviet officials con-
sider the development of decision-making ma-
chines to be essential to the successful man-
agement of their Acrea.singly cPmPlex gC9-
nomic and social system, and are giving sub-
stantial support and Party encouragement to
artificial-intelligence studies.
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Soviet ideologists and scientists are discuss-
ing, from a wide range of viewpoints, the
fundamental philosophical problems about
the nature of intelligence and of man which
are raised by attempts to model decision-
making or intelligent behavior. The stric-
tures of dialectical materialist dogma are not
inhibiting research in this area. In view of
the high prestige of the Soviet scientists tak-
ing part in these discussions and the publica-
tion in the Soviet scientific 'literature of the
opinions expressed, artificial-intelligence stud-
ies represent rare and significant examples
of intellectual freedom in the USSR.
Current Soviet work in the major subfields
of artificial-intelligence research includes in-
vestigation of techniques for machine search
of problem solutions, pattern recognition, ma-
chine learning, planning and induction in
machine systems, and brain modeling. Soviet
progress in research on machine techniques
for search, pattern recognition, and learning
compares favorably with US advances. Sci-
entists and engineers engaged in research on
pattern-recognition in the Soviet Union have
made original contributions to the theoreti-
cal basis for artificial pattern-recognition sys-
tems and have fabricated various types of
pattern-recognition devices.
Soviet research�On simulation of such as-
pects of the cognitive process as learning and
induction at present consists mainly of study
of self-teaching and self-organizing systems.
In these fields Soviet scientists and engineers
have created models based on neuro-physio-
logical conditioned-reflex approaches, on psy-
chological learning theory, and on automatic
control system theory.
Soviet study of the exceedingly complex
problems involved in incorporating techniques
of planning and induction in artificial prob-
lem-solvers is at a very early stage of devel-
opment. As yet, little progress has been made
in these fields. in the West.
Research on the modeling of brain processes
is receiving much emphasis in the USSR.
Since 1955, laboratory experimentation on
models that simulate neural processes has in-
creased, and outstanding neuro-physiologists
and psychologists have begun to work closely
with mathematicians, computer specialists,
and electronics engineers on cybernetic stud-
ies of brain-modeling problems.
Seminars are regularly offered to expose
students to the state of the art in the fields
that are especially significant to the future
of artificial-intelligence research in the USSR.
These seminars are conducted by the leading
researchers in artcial-iiifelligence tind are
thereby laying a sound foundation for future
advances in a new and multidisciplinary field.
DISCUSSION
INTRODUCTION
Since the mid-1950's, the Soviet govern-
ment has increasingly emphasized the impor-
tance of a cybernetics research program for
achieving national and international goals.-
A large part of the theoretical research aspect
of this program has been focused on the de-
velopment of decision-making machines.
Realization cif stithinachihes'would assist the
Soviets in solving two types of fundamental
�AEC. 828/46, "The Meaning of Cybernetics
the USSR," Cybernetics in the USSR, 30 Mar 84
problems: (1) decision making on the basis of
incomplete data (sometimes referred to as de-
cision making under conditions of uncer-
tainty), and (ii) decision making in the pres-
ence of complete but overwhelming quantities
of data. Problems of the latter type are by
definition beyond the capabilities of human
decision makers. Problems of the first type,
while inherently__ characteristic of human
thinking activity, are rapidly extending
beyond the bounds of possible human solution
in the context of economic management and
state administration.
A ,
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
ing systems.
id engineers
euro-physio-
hes, on psy-
automatic
ly complex
, techniques
iiicial prob-
N of devel-
been made
n processes
the USSR.
itation on
ses has in-
tysiologists
ork closely
specialists,
ietic stud-
to expose
the fields
he future
the USSR.
le leading
and are
or future
lary field.
basis of
to as de-
uncer-
aie pres-
tan tities
are by
human
�st type,
human
tending
5olution
:tit and
Decision-making machines are, in essence,
models of one variety or another of the human
problem-solving process. Models of the hu-
man thinking or problem-solving process are
artificial simulations of the information flow,
or information processing, procedures of the
human problem solver. The model may be
mathematical, in which case it may be a pro-
gram for a universal (digital) computing ma-
chine, or it may be a piece of hardware, such
as an electronic circuitry analog. (Many de-
cision procedures already elaborated by arti-
ficial-intelligence reseirch are not efficiently
modeled by the single-track, sequential meth-
od of operation of conventional digital com-
puters. This would not be surprising in view
of the apparent parallel operation of infor-
mation processing in the human brain. Such
decision routines can be modeled for demon-
stration purposes on a digital computer, but
analog models of some sort will loom increas-
ingly important in the fabrication of artificial-
decision makers.) In any case, the essential
feature of such a model is that for a given
input of information the model shall produce
at least the same output (of information) as
does the natural process being modeled. Re-
search on such models is categorized in the
United States under the rubric "artificial-in-
telligence."
Soviet reseltrch -eh-artificial-intelligence has
mushroomed since 1956 along with other sub-
problems of cybernetics. After the establish-
ment in 1959 of a national program for cy-
bernetics, the direction of Soviet research on
artificial-intelligence became one of the re-
sponsibilities of the Scientific Problem Coun-
cil on Cybernetics of the Presidium, Academy
of Sciences, USSR. The Cybernetic Machines
Section of the Council appears to be the unit
which assigns, monitors, and integrates most
of this research work on a national scale.
Actual research is conducted at the labora-
tories and institutes listed in the appendix.
The number of facilities engaged in this re-
search can be expected to grow as the cyber-
netic approach continues to permeate other
areas of Soviet science and technology appli-
cable to communication and control problems
arising in production industries, the economy,
law, government administration, and military
activities.
Although much of the application-oriented
cybernetics research is at present directed to-
ward the realization of more sophisticated
conventional systems for automatic control
of machines, plants, space research vehicles,
or even economic units, a large part of the
theoretical research is related, directly or in-
directly, to the development of "thinking ma-
chines," or "thinking cybernatons," as Soviet
scientists often refer to them. Realization of
such machines will have immediate relevance
to information abstracting and retrieval, ma-
chine translation, and general problem solv-
ing, and eventual application to later-genera-
tion automatic control systems.
A. A. Lyapunov, a leading Soviet mathema-
tician and editor of Problems of Cybernetics,
has described the relationship of artificial-in-
telligence to general cybernetics and the na-
ture of future control systems. He suggests
that algorithmization (mathematical model-
ing) of thinking processes and of control proc-
esses is one of the most important aspects of
cybernetic theory.' Other members of the
Scientific Council on Cybernetic., in explain-
ing the importance of artificial-intelligence,
state that "the basic products of radioelec-
tronics are variotis deviCeslar- automatic reg-
ulation, monitoring, control, and communica-
tions. The brain, which fulfills these func-
tions in the living organism, works much more
reliably and productively than any present-
day radioelectronic machine. . . . Some first
steps have already been taken in the direction
of constructing practical automata analo-
gous to the brain." 2
PHILOSOPHICAL ASPECTS
Reexamination of basic philosophical atti-
tudes toward the nature of the brain, of mind,
and of man himself is taking place wherever
artificial-intelligence research is being con-
ducted. The Soviet Union is no exception.
Philosophical-theOtetical queitiOns-bearing on
the simulation of intelligence, or intelligent
behavior, are being thoroughly aired and in- 3
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
vestigated in Soviet scientific circles, outside
as well as within the context of the funda-
mentals of dialectical materialism.2-8
A highly significant discussion took place
during June 1962 at a conference in Moscow
on "The Philosophical Problems of Cyber-
netics." This meeting was co-sponsored by
�the Scientific Council on the Complex Prob-
lem "Philosophical Questions of Natural Sci-
ence," the Scientific Council on Cybernetics,
and the Party Committee, all of the Presidium
of the Academy of Sciences, -USSR. It was
attended by about 1,000 specialists, including
mathematicians, philosophers, physicists, en-
gineers, biologists, medical scientists, lin-
guists, psychologists, and economists, from
many cities of the USSR.3 Participants in-
cluded such prominent cyberneticians* as
A. N. Kolmogorov, V. M. Glushkov, A. I. Berg,
P. K. Anokhin, A. V. Napalkov, A. A. Feld-
baum, A. A. Lyapunov, S. V. Yablonskiy, I. B.
Novik, and Yu. Ya. l3azilevskiy. Of the 10
reports presented, six were closely related to
the problem of artificial-intelligence.' A re-
port of this conference stated that "the prob-
lem most animatedly discussed was the most
disturbing of all, the problem of the techno-
logical operation of complicated psychic proc-
esses�the problem most frequently desig-
nated by the short and convenient although
not, in our view, entirely-correct-formula of
'can a machine think?'"
The discussion brought out three questions
as approaches to this problem. First, Is it
possible in general to reproduce with models
the complex intellectual activity of man?
Second, Can a machine surpass man in the
realm of intellectual activity and particularly
in the performance of creative tasks, such as
the formulation of new problems? Third, Is
It possible in principle, to achieve the exist-
ence of consciousness in a machine similar
to that exhibited by man?
Discussion at the June conference of the
first question may be described as informa-
tional and confirmatory rather than argu-
� The term "cybernetician" is�applied by the
Soviets to those practicioners of traditional disci-
plines who exhibit the common characteristic of
dealing with their research problems in the frame-
work and methodology of cybernetics.
mentative. There was general agreement
that, in view of the replications of intellectual
and sensory functions of man already
achieved, an affirmative answer to questions
about the reproduction of human cognitive
processes cannot be doubted. The second
question was discussed more vigorously. Life
scientists, represented by P. K. Anokhin, a
leading neurophysiologist, argued that the
potentialities of a machine are limited to solv-
ing problems assigned by man using algo-
rithms of decision which man puts into the
machine. This position was countered by
V. M. Glushkov, a leading Ukrainian cyber-
netician, and A. A. Feldbaum, Doctor of Tech-
nical Science in Electronics and member of
the automation faculty of the Lenin Power
Institute. They pointed out that there are
machines today which, in the process of solv-
ing one complicated problem, can independ-
ently pose and solve a series of autonomous
problems of a particular character. Glush-
kov maintained that any form of human
thought can, in an informational model, be
reproduced in artificially created cybernetic
systems. He agreed, however, that by virtue
of historical necessity�the fact that it was
precisely man who created machines for his
use and not the reverse�the destiny of man
will always be the more important in the
processes of thought and cognition:- Thus
Glushkov's assessment is that a machine can
be "smarter" than one man, or even a group,
but it cannot be "smarter" than human so-
ciety as a whole.
Questions about machine consciousness
were argued most sharply at the conference.
One group argued the "black box" approach
to solution of these questions. This viewpoint
holds that man judges the presence of con-
sciousness in other people by analogy, i.e., by
observing the behavior manifested by others
in response to inputs (stimuli) and by com-
paring such manifestations with his own be-
havioral responses of which he is consciously
aware. Therefore, if a machine faithfully
produces the same outputs as a man in re-
sponse to the same inputs, it can be analo-
gously inferred to possess consciousness ac-
cording to this view. On the other side it was ti
maintai
conscio
elemen
result
which �
social e
� A m
nem wa
scientis
which e
the, bra
the hu
nologic
sibility
to be
Kolmog
is opti
brain-m
freeing
arbitra
these c
Kolmog
point o
scribed
replica
can be
element
have th.
system
rov con
As
can in
the m
It is
within
do not
argum
quest!
porary
ural o
consci
The
the Ko
ascenda
view on
have be
the W
somewh
- philoso
The gro
argume
a think
Approved for Release: 2018/03/28 C06731721
agreement
intellectual
already
questions
cognitive
e second
usly. Life
okh in, a
that the
ted to solv-
sing algo-
into the
ntered by
ian cyber-
r of Tech-
ember of
nin Power
there are
of solv-
independ-
tonomous
r. Glush-
f human
model, be
ybernetic
by virtue
at it was
es for his
y of man
t in the
n. Thus
me can
a group,
uman so-
iousness
ference.
pproach
'ewpoint
of con-
i.e., by
others
by corn-
own be-
sciously
aithfully
in re-
analo-
ness ac-
le it was
Approved for Release: 2018/03/28 C06731721
maintained by some at the conference that
consciousness necessarily includes a subjective
element with a specific character which is the
result of the labor and social relations in
which people engaged during the process of
social evolution.
A more empirical approach to conscious-
ness was advanced at the conference by those
scientists interested in modeling the structure
which embodies human consciousness, that is,
the brain. Although a structural model of
the human brain is .far beyond present tech-
nological and scientific capabilities, the pos-
sibility of its future realization is considered
to be worth discussion by the Soviets. A. N.
Kolmogorov, world renowned mathematician,
is optimistic about the chances of success in
brain-modeling ventures. He called for the
freeing of definitions of life and thought from
arbitrary premises and for the redefinition of
these concepts along purely functional lines.
Kolmogorov believes that if the functional
point of view toward life and thought is sub-
scribed to, the conclusion is inevitable that
replication of the organization of a system
can be accomplished by organizing different
elements into a new system which would
have the essential traits and structure of the
system being modeled. From this Kolmogo-
rov concluded that:
A sufficiently complete model of a living being
can in all jiistice- becalled a living being and
the model of a thinking being, a thinking being.
It is important distinctly to understand that
within the limits of a materialistic ideology there
do not exist any kind of well-grounded, principal
arguments against a positive answer to [this]
question. This positive answer is the contem-
porary form of the attitude concerning the nat-
ural origin of life and the material basis of
consciousness.
The extreme empirical view advanced by
the Kolmogorov school seems to be in the
ascendancy. For example, several points of
view on the question, "Can a machine think?"
have been reviewed by a group of authors in
the Works of the Kazan Aviation Institute, a
somewhat surprising source for discussions of
philosophical aspects of artificial-intelligence.d
The group did not find convincing any of the
arguments against the possibility of creating
a thinking machine. The subject of the neg-
-
ative arguments considered by them ran the
gamut from the algorithmic insolvability of
some problems, and the irreducibility of the
thinking process to a physical operation, to
the impossibility of modeling the subjective-
psychological world of man. The Kazan
group points out in rejecting such arguments,
that cybernetics provides the first basis for
(i) uncovering the elements involved in con-
sciousness and cognition and (ii) "resolving
positively the question of the possibility of
creating a thinking machine." The Kazan
group does not recognize the brain as the only
highly organized material system in which
consciousness can develop, and forecasts that
highly organized material systems of another
type, in which consciousness develops, are pos-
sible and realizable.
V, M. Glushkov, one of the most politically
powerful among Soviet cyberneticists,* sides
with Kolmogorov and the Kazan group. Re-
cently, in discussing the possibilities of ma-
chine intelligence, he contrasted cybernetic
systems with earlier mathema.tico-logical and
other formal language approaches to model-
ing the thought processes. He found signifi-
cant advances in the cybernetic approach.
Glushkov would describe any control or cogni-
tive system as a cybernetic system which can
be analyzed as an abstract [informational]
model. For this purpose, both the input and
output information, that is, .all of- the _infor- _
!nation which a system exchanges with the
outside world, can be conceived as being en-
coded in words of a given standard alphabet.
All of the activity of the cybernetic system
may thus be reduced to the transformation of
words in a standard alphabet. The study of
a given cybernetic system can be reduced
thusly to the determination of rules accord-
ing to which the indicated transformation
occurs. Glushkov noted that among these
rules there may be some which permit certain
V. M. Glushkov is Vice President of the Academy
Of Sciences, Ukrainian Soviet Socialist Republic;
Director of the Institute of Cybernetics in Kiev;
Chairman of the Cybernetics Council of the Ukrain-
ian Academy; and Chairman of the recently created
Interdepartmental Council for the Introduction of
Mathematical Methods and Electronic Computers
in the National EcOnomy, which is under the State
Committee for Coordination of Scientific Research
of the Council of Ministers, USSR.
Approved for for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
chance transformations, as well as some that
permit the altering of other rules of informa-
tion conversion in the course of time under
the influence of an infinite surrounding me-
dium. Appreciating the possible infinitude of
the system of rules defining the regularities
of the informational activity of the brain,
Glushkov believes that the modeling of a suffi-
ciently large number of the essential rules in
the brain system eventually will result in a be-
havior pattern of the model (on the informa-
tional level) which will correspond to brain
activity!' Glushkov has asserted that, since
modern electronic digital computers possess
"algorithmic universality� . . . it is theoreti-
cally possible to model [on the informational
level] any thought processes with the aid of
such machines." "- P 10
At least as significant as the content of
these epistomological disputations is their
wide distribution in popular and Party media.
Typical of "thinking machine" discussions is
an article in Znaniye-Sila (Knowledge is
Power, a popular-science type of magazine).
It describes the parallel mode of operation of
the brain (carrying out several calculations
or decision procedures simultaneously) versus
the series (single-track) nature of contem-
porary electronic computing _apparatus and
points out the advantage of the former in effi-
ciency and universality. It reports that So-
viet researchers are studying the operating
principles of automata which work in parallel
instead of in series.9
Soviet research in artificial-intelligence is
motivated by the anticipated necessity of
using machines in place of people in situations
where speed, complexity, or other character-
istics of control processes exceed the capa-
bility of man. The Soviet policy regarding
the use of "thinking machines" was expressed
in a recent edition of Kommunist. The de-
velopment of technology, with the increase in
speed and accuracy requirements of separate
production operations and the growth of the
'That is, computers can perform any information
transformation on the basis of a program .(algo-
rithm) built out of their available elementary in-
structions, if these include rules which dellnechance
transformations and instructions ,by which certain
alterations are made in the system of rules.
entire technological process as a whole, was
said to have begun to exceed, in most cases,
man's power to control them. The Kommu-
fist article concludes that it is necessary to
replace even the psychic activity of man in
such cases with automatic control machines.�
PRINCIPAL RESEARCH PROBLEMS
The "tent-like" character of artificial-intel-
ligence research has resulted in a somewhat
chaotic state in this field of science."-I3
There are several schools, each represented
by spokesmen as critical of other schools as
they are competent in the techniques of their
own. The result is a lack of standardized cri-
teria for use in assessing Soviet research in
the field of artificial-intelligence. Thus, an
expert in one popular US approach, upon dis-
covering a lack of comparable work in the
USSR, will give a negative evaluation of So-
viet research in artificial-intelligence. On the
other hand, the opponents of that particular
US expert will argue that the absence of such
research in the USSR signifies that the So-
viets have withdrawn from a blind alley of
investigation. Many of these conflicts are
semantic; almost all of the approaches to
artificial-intelligence share a common set of
problems. When Soviet research on �these -
problems is compared with US approaches to
the same problems, regardless of "schools,"
the USSR and the US are found to be approxi-
mately on a par.
There are differences in emphases, however,
between Soviet and US approaches to these
shared problems. US scientists tend to em-
phasize mathematical or machine models of
human cognitive processes. Many Soviet re-
searchers on the other hand consider that the
human brain has reached certain limits in re-
gard to its capabilities for memory and oper-
ational speed after a long process of slow
evolutionary development." The resultant
qualities of the human brain, therefore, are
not believed by the Soviets to be equal -to all -
the tasks modern men must face. According
to them, a machine that might be a perfect
analog of the brain, for instance, will not do
any better than the brain when faced with (o
such t
plex p
"brains
built.
emy of
asserts:
technol(
with fu
nology
will giv
In nat
copied
gram a
US sci
artifici
tern r
duction.
Search
The
problem
tempor
USSR
and er
solutio
problem
non-triv,
solutio
method
suming
Soviet
duction
some r
cal mac
the int
niques.
pattern-
sify pro
tin ty.
Soviet
such tec
more an
Theor
devices
. 1953."-2
psychol
sively
tions of
cently, 1
Approved for Release: 2018/03/28 C06731721
_Approved for Release: 2018/03/28 C06731721
whole, was
ost cases,
e Kommu-
ecessary to
of man in
achines."
EMS
cial-intel-
somewhat
ience."-'3
epresented
schools as
es of their
dized cri-
� earch in
Thus, an
, upon dis-
krk in the
ion of So-
:. On the
particular
ce of such
1 the So-
1 alley of
filets are
'aches to
on set of
on these
oaches to
'schools,"
approxi-
however,
to these
d to em-
aodels of
foviet re-
that the
its in re-
nd oper-
of slow
esultant
'ore, are
al to all
:cording
perfect
not do
ed with
such tasks. Therefore, if the ever more com-
plex problems are to be solved, machine
"brains" surpassing those of men must be
built. M. V. Keldysh, President of the Acad-
emy of Sciences, USSR, in making this point
asserts: "We must copy nature's processes in
technology creatively rather than literally,
with full knowledge of nature and of tech-
nology so that we may select techniques which
will give us better results than those achieved
In nature." 's The natural processes to be
copied "creatively" in the Soviet research pro-
gram are the same processes investigated by
US scientists of most schools concerned with
artificial-intelligence. These are search, pat-
tern recognition, learning, planning, and in-
duction."
Search and Pattern Recognition
The first approach to machine solution of a
problem is search. Given a problem, a con-
temporary data-processing machine in the
USSR as elsewhere, can search rapidly by trial
and error through a large number of possible
solutions for a valid solution to the given
problem. Nevertheless, in solving complex
non-trivial problems, the number of possible
solutions is so large that this trial-and-error
methodology becomes excessively time-con-
suming in practical operation.
Soviet scientists recognize- that a large re-
duction in search time, although bringing
some real problems into the realm of practi-
cal machine solution, could be achieved by
the introduction of pattern-recognition tech-
niques. The machine that is designed with
pattern-recognition aids-to-search could clas-
sify problems into categories amenable to cer-
tain types of solutions. The current state of
Soviet and Western research suggests that
such techniques are nearing practicability for
more and more complex patterns.
Theoretical studies for pattern-recognition
devices began in the USSR as early as
1953.'7-2" The philosophical, physical, and
psychological bases of perception were exten-
sively discussed in Voprosy Psikhologii (Ques-
tions of Psychology) in 1959." More re-
cently, Soviet researchers have accomplished
considerable work on the specifics of mini-
mum descriptions of images that are required
for recognition by artificial systems.22-27
The works of E. L. Blokh, mathematician
at the Institute for Information Transmission
Systems and E. M. Braverman of the Insti-
tute of Automation and Telemechanics are
notable among recent approaches to practical
solutions of recognition problems. Braver-
man has originated a "compactness hypothe-
sis" � as a theoretical basis for solution of
such problems. Several Soviet researchers
are using this, theory as a basis for develop-
ment of specific recognition techniques. E. L.
Blokh is using certain operations to compute
the distance and angle between elements, rep-
resenting various patterns presented, in an
n-dimensional configuration space. This ap-
pears to be a promising mathematical model-
ing approach to a large class of pattern-rec-
ognition problems."
Pattern-recognition modeling studies are
being supported by research on perception
from the psychological and physiological
points of view. One investigation involved
the establishment and development of percep-
tual activity in 3- to 6-year-old children. Eye
movements' of the subjects were observed and
recorded photographically as the children ex-
amined (for learning) and later recognized
pictures presented to them. The -eye-move-
ments recorded were then compared with
measures of the recognition ability of the chil-
dren at different ages.s� Another study was
devoted to identification of objects in the
visual system. Time required for subjects to
recognize simple objects, formed of small
numbers of elements, correlated well with
an information theory model for such recog-
nition that was based on earlier findings on
information Collating and processing activity
in the eye system."
The "compactness hypothesis" is formulated as
follow's: Patterns presented to the artificial system
are characterized by a number of criteria equal to
n. The values of the n criteria are used to estab-
lish points in an n-dimenSional configuration space.
Each point represents an individual pattern. All
points representing 'Patterns which are similar, for
example, all letters "A," all figures "5," all pictures"
of cats, will tend to lie in "compact" regions of the
space, with relatively easily discernible separations
between regions.v'
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
Soviet work in the area of realization of
physical models to perform pattern recogni-
tion has been reported. At a 1957 Scientific-
Technical Conference on Cybernetics, one
reading device for an information machine
was described." At an All-Union Conference
on Machine Translation, held at Moscow State
University in May 1958, reports on principles
of constructing electronic reading devices
were heard, and a device "enabling the blind
to_read ordinary typographic text" was de-
scribed.33 34 (The latter device has been pic-
tured in the Soviet press, but its operating
principles were not described in publication.)
Hardware modeling of pattern-recognition
schemes is being conducted by A. A. Kharke-
vich, the well-known radio engineer, as well
as others.33-37 51 Some of this research is di-
rected toward the specific application of auto-
mating the input of information into comput-
ing machines.-41
A quasi-topological method of distinguish-
ing and identifying letters has been developed
and realized in hardware by a group of re-
searchers of the Institute for Systems for
Transmission of Information (Moscow) .42
This makes use of scanning the contour of a
letter with a light spot and identifying its
topological characteristics (ends of lines, and
lunetions- of lines), which are recorded in a
binary code. Since this will not separate all
Cyrillic letters, some of which are topologi-
cally identical, further geometric analysis is
used to analyze topologically redundant
groups. Such a 'scheme is basically sound in
theory and relatively easily realized in hard-
ware, but system "noise!' (disconnected lines
or smudged letters) may be hard to deal with,
and no figures have been published on reli-
ability of recognition accomplished.
In June 1960, the Scientific Council on Cy-
bernetics sponsored a seminar on reading de-
vices. This seminar considered the principles
of constructing such machines and creating
corresponding systems for coding the infor-
mation involved." Five different machines
under development for automatic pattern
recognition were described by V. M. Cilushkov
(letter recognition by line scan and minimal
description); V. A. Kovalevskiy (image scan-
ning by following the outlines of letters);
A. D. Krisilov (identification of constant fea-
tures of letters by means of standard tele-
vision techniques); V. M. Tsirlin (the quasi-
topological method) and A. G. Vitushkin (a
computer manipulated system for analyzing
Cyrillic letters which separates characteristic
features by means of vertical line scanning).
In addition, E. M. Braverman and V. S. Fayn
presented papers on recognition systems em-
ploying learning (that is, performing identi-
fication on the basis of criteria not given be-
forehand).
Studies on the mathematical modeling of
recognition processes on electronic digital
computers have been conducted by M. M.
Bongard, an outstanding young biophysi-
cist.43-" A report of his in the Cybernetics
Council collection Biologicheskiye Aspekty Ki-
bernetiki�Sbornik Robot (Biological Aspects
of Cybernetics�Collected Works) describes
the methodology and prospects of recently
begun research aimed, first, at bridging the
gap between physiological study of optical re-
ceptor activity and the modeling of recogni-
tion, using a universal digital computer.�
However, Bongard alludes to the disadvantage
of this inodel in contrast to the parallel in-
formation processing employed in human rec-
ognition activity. He foresees, therefore, the
development of a "logic of recognition . . .
a logic such that it could be used in an analog
computer. In essence, such a machine will
be a model of part of the human brain."
Principles for constructing a "universal
reading" machine have been developed by
V. M. Glushkov." This scheme uses a cath-
ode ray tube-type receptor, a computer to
control the trace and to compute the descrip-
tion of the pattern presented, and a tech-
nique of comparison against descriptions pre-
stored in the memory for identifying patterns
,presented. The author admits that the
scheme described is unnecessarily cumbersome
for the recognition of such simple stylized
patterns as digits or letters, but points to its
usefulness for "reading" complex contours or s
semitc
ing at
putini
Ukrait
with a
A sc
has be
Comp
didate
skiy."
recogn
with ;
printeu
using
to Kol
bon cc
each
latter
transf(
many
artifici
and it
incorn
For
ognizit
algorit
dichot(
set of
skiy be
it poss
ing m
recogn
ory ca
Learni
run
solving
the ad
;:machil
Its alr.
-191 prot
.7.4a3m3 p
peri(
a �
plar
an
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
r (image scan-
?.s of letters);
r constant fea-
standard tele-
in (the quasi-
Vitushkin (a
for analyzing
characteristic
ne scanning).
nd V. S. Fayn
systems em-
ming identi-
not given be-
modeling of
Tonic digital
!d by M. M.
ng biophysi-
?. Cybernetics
e Aspekty Ki-
gical Aspects
:s) describes
t of recently
bridging the
of optical re-
of. recogni-
computer."
ilisadvantage
parallel in-
human rec-
erefore, the
tnition . . .
an analog
whine will
rain." '
"universal
eloped by
es a cath-
nputer to
e descrip-
1 a tech-
;ions pre-
patterns
hat the
tbersome
stylized
Ls to its
tours or
semitone pictures. Such a "universal read-
ing automaton" has been built at the Com-
puting Center of the Academy of Sciences,
Ukrainian SSR, and is used in conjunction
with a "Kiev" computer.
A somewhat different "machine that reads"
has been developed at the Ukrainian Academy
Computer Center under the direction of Can-
didate of Technical Sciences, V. A. Kovalev-
skiy." 50 This system reportedly failed to
recognize only 2 of 35,000 numbers produced
with a portable typewriter, including half
printed and otherwise distorted samples. In
using this method for recognition, according
to Kovalevskiy, the maximum of the correla-
tion coefficient for an unknown image and of
each of the standard images is sought, the
latter images being subjected to all possible
transformations. When this is done, all nor-
mally typed letters, as well as most of those
artificially marred, are correctly recognized
and identified, with the statistical error of
incorrect recognition not exceeding l0-4.
For solving a more general problem of rec-
ognizing nonstandard letters and numbers, an
algorithm is being developed which is based on
dichotomy, that is, sequential division of the
set of all images into two classes. Kovalev-
skiy believes that such an algorithm will make
it possible to work with an alphabet contain-
ing many characters and will assure rapid
recognition with a comparatively small mem-
ory capacity.
Learning, Induction, and Planning
Further improvement in machine problem-
solving efficiency could be accomplished with
the addition of a learning capability. The
machine would then be able to apply readily
its already proven methods to the solution
of problems that are new but similar to prob-
lems previously encountered in the machine's
experience. Radical reductions of search
time could be realized through the application
of planning methods: the machine would sur-
vey ,and analyze the solution space and plan
the best way for its detailed examination.
Furthermore, to manage broad classes of very
complex problems, the machine, as the
human, must construct and internalize a
model of its environment, that is, it must
employ some scheme of induction.
Research on planning and induction in ar-
tificial systems is at a rather early stage of
development in the USSR, as it is in the West.
Progress is occurring, however, in fields con-
tributing to the development of machine
learning, induction and planning. Such sup-
porting research includes studies on informa-
tion theory, coding theory, brain modeling,
statistical decision theory, automata theory,
and heuristic programming. Pertinent So-
viet literature often treats these subjects as
conjoined in such studies as pattern recogni-
tion employing learning, other learning sys-
tems, self-organizing systems, or brain models.
Learning appears to be an essential charac-
teristic of more efficient and truly universal
pattern-recognition systems, just as it is of
more efficient problem-solving apparata in
general. Soviet researchers in the field of
learning systems like Braverman, Glushkov,
and Mark Ayzerman,* compare favorably
with their Western counterparts. Further-
more, they are working on essentially the
same types of studies: perceptron-type sys-
tems, algorithms for teaching the recognition
of shapes, and computer programs for recog-
nition of pattern configurations." 51-03
Investigations of learning 'systems for rec-
ognition are being conducted at a variety of
Soviet scientific establishments. At the In-
stitute of Automation ,and Telemechanics in
Moscow, a machine was programmed to dis-
cern numerical figures written in different
handwritings. According to Soviet reporters,
in only a few cases did the machine give er-
roneous responses, even when confronted with
previously unseen figures. The Institute of
Surgery of the Academy of Medical Sciences
is testing the hypothesis that a "compact
area" is formed in the brain of an animal or
a human by variants of a similar image. The
Institute of Biophysics of the Academy of Sci-
�M. A. Ayzerman Is a Doctor of Technical Sciences
in mathematics and electronics at the Institute of
Automation and Telemechanics, Moscow.
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
ences is attempting to program a machine
that will identify indices of an image and,
on the basis of the indices, to recognize the
image. In each of the experiments, errors
were made by the machine. However, the So-
viets believe that the important fact is that
the machine is capable of accumulating ex-
perience, with the result that its qualifications
are increasing and its errors are gradually
decreasing."
M. A. Ayzerman's latest experiments involve
the teaching of a machine to recognize pat-
terns without a need for pre-introduced cri-
teria. Starting from a basis in Braverman's
"compactness hypothesis," Ayzerman devel-
ops two algorithms by which a machine can
"learn" to distinguish between the compact
areas representing different images in a con-
figuration space, thus separating and recog-
nizing the images. The first algorithm has
the machine construct, one by one,, random
hyperplanes whose only criterion is that they
separate points which the machine is told
(that is, training process) represent different
objects. Ending up the training phase with
a series of intersecting hyperplanes, the ma-
chine then examines these and "washes out"
sections of planes which do not perform the
separating function, thereby leaving a series
of broken hyperplanes which effectively sepa-
rate areas containing points representing dif-
ferent images. In using .the second algo-
rithin, the machine constructs positive poten-
tial surfaces (functions) decreasing away
from, respectively, each point or set of points
representing images of the same object.
Identification of a test image (point) is ac-
complished by (first algorithm) determining
on which side of the hyperplanes the point
lies, or (second algorithm) determining which
potential surface (function) has the highest
value at the test point. Tests were conducted
with five digits (0, 1, 2, 3, 5), each written
160 different ways. Using 40 samples of each
digit for training, and 120 for test, the ma-
chine achieved 83-89 percent correct recog-
nition with the first algorithm. With "paral-
leling" of seven variations of a digit in the
training process, 98 percent correct response
was achieved. The training sequence filled
1,500-3,000 binary digits in machine memory.
The second algorithm was tested using 10
samples of each digit for training and 150
for testing and resulted in 100 percent cor-
rect recognizance. Additionally, the second
algorithm was tested on the 10 digits from 0
to 9, with 10 samples for training and 140 for
test on each digit, and achieved 85 percent
correct response.28"57 Ayzerman's second
algorithm is very similar to that employed in
a US device now becoming operational for the
identification of sonar contacts.
Pattern-recognition techniques are em-
ployed at the Institute of Surgery imeni Vish-
nevskii, Moscow, to achieve rapid assessment
of the area and seriousness of burns." The
algorithm, for recognition of objects with
many parameters, employs learning. The
system is "trained" on case histories. When
vital information, such as burn area and lo-
cation and patient's age, is fed in, the com-
puter identifies and stores symptoms and
other factors. It also identifies objective cri-
teria for forecasting the outcome of the ill-
ness. The system was tested on additional
case histories with known outcomes, and the
prognoses in ,most cases agreed with the ac-
tual course and outcome of the injury.
Many of the Soviet attempts to realize mod-
els of learning, induction, and other aspects
Of the cognitive process- are carried on under
the classification of self-teaching (or learn-
ing) systems, or of self-organizing systems.
A significant portion of research in these areas
has apparently not been published. In a
number of cases the titles of papers discussed
at meetings and seminars have been pub-
lished, but the contents of the papers are un-
available to the West. Thus, a self-teaching
machine based on a program model was dis-
cussed at the First All-Union Meeting on Com-
puter Mathematics and Computing Technol-
ogy held in 1959, but details have not been
circulated to the West. Other self-teaching
machines were alluded to (but only in gen-
eral terms) at cybernetics seminars at Kiev
and Moscow State Universities."-67 Since its
inception in 1955, the latter seminar, con-
%.1
duc
Lya
out
ject:
Ai
cont
stud
knoN
A. G
gori;
learz
ing)
work
stud.
variE
trol
(self
appl:
is, tt
mula
Ivakl
ceptr
in tb
ceptr
ation
lar U
Mo
autor
collec
trol.
Thor]
by, t
learn.
Sovie
and
Newe;
Roche
Andre
Mosec
Soviet
algori
finds
rithm
deteri
put ir
ing pi
tern.-
Clo;
learni
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
chine memory.
sted using 10
ining and 150
0 percent cor-
ly, the second
digits from 0
ng and 140 for
ied 85 percent
man's second
at employed in
�ational for the
r.
rues are em-
!ry imeni Vish-
pid assessment
lourns.84 The
objects with
learning. The
stories. When
n area and la
d in, the corn-
iymptoms and
z objective cri-
ome of the ill-
on additional
comes, and the
xi with the ac-
: inj ury.
to realize mod-
other aspects
Lrried on under
Ling (or learn-
nizing systems.
in these areas
Lblished. In a
epers discussed
awe been _pub-
papers are un-
a self-teaching
model was dis-
leeting on Com-
ruting Technol-
have not been
er self-teaching
at only in gen-
minars at Kiev
s.88-87 Since its
� seminar, con-
ducted by the outstanding cyberneticist A. A.
Lyapunov, has held biweekly sessions through-
out the school year on a wide range of sub-
jects related to cybernetics.
Application of principles from automatic
control theory to self-organizing systems
study is exemplified by the work of the well-
known control engineer and mathematician,
A. G. Ivakhnenko. He has surveyed and cate-
gorized various types of "learning" and "self-
learning" (that is, new information generat-
ing) systems and has related US to Soviet
work on some of these types. Ivakhnenko is
studying the application of the theory of in-
variance and the principles of combined con-
trol systems to the development of certain
(self) learning systems. Control principles
apply to the memory part of the system, that
is, to the control of the selection and accu-
mulation of information in the memory.
Ivakhnenko is specifically interested in a per-
ceptron-type device, a scheme first developed
in the United States.6688 Ivakhnenko's per-
ceptron device apparently employs some vari-
ations and innovations in comparison to simi-
lar US devices."
Modeling the processes of instruction with
autornatic systems was discussed in a 1962
collection on automatic regulation and con-
trol. Starting from the learning theories of
Thorndike, Gestalt psychology and I. P. Pav-
lov, the authors discussed various machine
learning systems. Among the Western and
Soviet systems discussed were the perceptron
and the approaches of the US scientists
Newell, Simon and Shaw, Gelernter and
Rochester, and 0. Selfridge, the UK scientist,
Andrew, and the Special Design Bureau of
Moscow Power Engineering Institute. The
Soviets view training as a process of changing
algorithms, and propose that "a system which
finds by means of automatic search an algo-
rithm of action which is successful from any
determined point of view and which was not
put into the system by man before the train-
ing process, should be called a learning sys-
tem." "
Closely related to systems embodying self-
learning are those capable of self-organiza-
Um.� Soviet scientists evinced interest in
the theory of self-organizing systems as early
as 1959. In that year, S. N. Braynes and A. V.
Napalkov wrote on the subject for Voprosy
Filosofii (Questions Of Philosophy). In that
study, the investigators related the develop-
ment of such systems to their work on con-
ditioned-reflex modeling. They foresaw the
realization of "an algorithm of operation for
self-organizing cybernetic systems, ensuring
the formation of new programs for operation
without the undertaking of 'exhaustive search'
of all possible variants.""
Considerable attention was devoted to self-
organizing systems at an All-Union Meeting
on Computer Mathematics and Computing
Technology (1959) and at a symposium on
Principles of Design of Self-Learning Systems
held in Kiev during 1961. Comparison of the
published papers from the latter symposium
with those given at the first US Interdisci-
plinary Conference on Self-Organizing Sys-
tems in 1959 reveals very similar topic cover-
age and a similar level of achievement re-
flected at the two conclaves. As of 1961, the
Soviets were 2 to 3 years behind the United
States in this particular area of artiflcial-in-
telligence research.74-78
Brain Modeling
Historically, there has been a large amount
of Russian neurophysiological research since
the early 19th century, but its mathematiciza-
tion is a recent innovation. Brain research �
now is very much concerned with the algo-
rithrnization and modeling of the information
transactions which take place in living or-
ganisms. These studies play an important
role in cybernetics/artificial-intelligence re-
�M. C. Yovlts, Chairman of the First and Second
Conferences on Self-Organizing Systems, Chicago,
1960 and 1962, considers these areas of artificial-
intelligence research to be of great significance . To
Yovits it appears that "certain types of problems,
mostly those involving inherently non-numerical
types of information, can be solved efficiently only
with the use of machines exhibiting a high degree
of learning or self-organizing capability. Examples
of problems of this type include automatic print
reading, speech recognition, pattern recognition, au-
tomatic language translation, information retrieval,
and control of large and, complex systems. Efficient
solutions to problems of these types will probably
require some combination of a fixed stored program
computer and a self-organizing machine," 72
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
search and are conjoined with attempts to
simulate artificially the pattern recognition,
learning, planning and induction processes.
Before research could begin in this new
field, the whole area of physiology and cyber-
netics had to be broken out of the restraints
of Pavlovian doctrine. The beginning of this
break was apparent in a 1955 review of the
subject by the well-known Soviet physiologist
P. K. Anokhin in Questions of Philosophy
(Voprosy Filosofii)." Laboratory experimen-
tation in the modeling of brain processes
began shortly thereafter. Some of the earli-
est work was reported at a Scientific-Techni-
cal Conference on Cybernetics, held at the
Laboratory of Electromodeling of the Acad-
emy of Sciences, USSR, in May 1957. L. I.
Gutenmakher, Director of the Laboratory,
described work on the electrical modeling of
certain mental work processes using "infor-
ma.tion machines with large internal stor-
age." 80 Research into the structural make-
up of the human brain was discussed at the
Seminar on Cybernetics at Moscow State Uni-
versity in 1960.8,
, After an artificial-intelligence slant to tra-
ditional neuroanatomy and neurophysiology
became evident in efforts to model the brain,
a new type of interdisciplinary scientist
emerged. A. V. Napalkov of the Faculty of
Higher. Nervous Activity at Moscow State Uni-
versity could well be described as the first
of this new breed of physiologist-cyberneti-
cist. In early 1959 he co-authored, with a
medical doctor and an engineer,a study which
surveys cybernetics and physiology in general,
including the theory of automata. Further-
more, these scientists describe the results of
studies on brain activity in terms of a search
for algorithms representing systems capable
of independent development of new programs
for their operation, and those able to form
new, behavior patterns on the basis of proc-
essing information accumulated earlier.
They also described an artificial device which,
in a primitive way, 'simulates these learning
processes, that is, a "learning automaton,"
and which was developed at the Moscow Power
Engineering Institute in cooperation with the
life scientists."
More intensive investigations into the infor-
mation processing procedures of the brain,
still in terms of the development of chains of
conditioned reflexes, were described by Napal-
kov in 1960.8" In 1962, the researchers in the
Department of Higher Nervous Activity re-
ported findings which showed increasing so-
phistication in the greater complexity of the
algorithms of information processing that
had been derived. By 1962, a much more so-
phisticated "learning machine," based on the
algorithms defined by the neurophysiologists,
and exhibiting some capability at "self-organ-
ization" (that is, self-improvement), had been
fabricated by the engineers at the Power En-
gineering Institute. This group worked with
the neurophysiological laboratory of S. N.
Braynes, co-worker and co-author with Napal-
kov."
Soviet Bloc researchers are also investigat-
ing the mode of operation of brain processes
from the point of view of psychology. The
work of a Czech, E. Goias, on the conditions
of generalization in pattern recognition and
learning falls into this class of research. His
experiments involved a statistical analysis of
the process of generalization as manifested
by subjects perceiving common elements
among sets of stimuli (objects) presented.
This study, clearly of a preliminary nature,
served only to demonstrate that wide varia-
tions characterize the conditions for general-
ization."
New centers for brain research along cyber-
netic lines are now being established at the
Brain Institute, Institute of Physiology, Insti-
tute for Information Transmission Problems
and at installations outside the Moscow-Len-
ingrad complex. At Kiev, for example, stu-
dents are offered the opportunity to obtain
training in the most advanced areas of arti-
ficial-intelligence research, and specifically
brain modeling. In 1962, two seminars were
held under the auspices of the Cybernetics
Council of the Ukrainian Academy of Sci-
ences. The first, on "Automation of Thinking
Processes," was conducted by V. M. Glushkov,
chairman of the Council. This seminar cov-
ered" (i) the foundations and particularities
of thought processes that are characteristic of
man it
the po ,
tion;
puters
tion, r�
meanir
coriseqi
strateg
music;
the In�
self -tea
merit
and (v
degree
esses ai
sirnulat
The 5
emy, w
N. M. A
cybernE
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
into the infor-
of the brain,
t of chains of
Fined by Napal-
oarchers in the
is Activity re-
increasing so-
oplexity of the
ocessing that
ouch more so-
based on the
physiologists,
d "self-organ-
zot) , had been
be Power En-
worked with
ary of S. N.
c with Napal-
so investigat-
sin processes
hology. The
le conditions
agnition and
search. His
LI analysis of
; manifested
on elements
) presented.
nary nature,
wide varia-
for general-
along cyber-
shed at the
ology,
in Problems
loscow-Len-
ample, stu-
F to obtain
eas of arti-
specifically
ainars were
Cybernetics
my of Sci-
g Thinking
Glushkov,
tninar coy-
ticularities
teristic of
man in the creative sphere of his activity, and
the possibilities of their algorithmic descrip-
tion; (ii) modeling on contemporary com-
puters of such processes as pattern recogni-
tion, recognition of concepts, identification of
meaningful sentences, deduction of logical
consequences, proving theorems, selections of
strategies in games, and composition of
music; (iii) learning as a basis for modeling
the mental activity of man; (iv) theory of
self-teaching systems and practical develop-
ment of algorithms incorporating learning;
and (v) correlations between precise (to the
degree possible) modeling of creative proc-
esses and the specifics of machine algorithms
simulating these processes.
The second seminar, at the Ukrainian Acad-
emy, was led by Doctor of Medical Sciences,
N. M. Amosov. Problems associated with bio-
cybernetics and the. application of electron-
ics in biology and medicine were considered
at this seminar. Specific topics included (i)
application of information theory in biology
and medicine; (ii) principles of automatic
control in biological systems and their peculi-
arities; (iii) some principles of coding infor-
mation in the nervous system; (iv) perception
and transformation in receptors and the cen-
tral nervous system; (v) contemporary hy-
potheses on the nature of nerve excitation
from the position of biocybernetics; (vi) some
questions of modeling elements of the central
nervous system; (vii) thinking and the psy-
chic activity of.man; (viii) principles of form-
ing self-organizing neuron nets and bionics;
(ix) control of the processes of excitation and
inhibition in the central nervous system by
means of electrical and electromagnetic in-
fluences; and (x) cellular biology in the light
of biocybernetics."
13
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
APPENDIX
Institutes and Scientists Associated with Research of
Artificial Intelligence Significance in the Soviet Bloc
Academy of Medical Sciences, USSR
Institute of the Brain:
Glezer, 'V. D. Retinal activity in identification
Nevskaya, A. V. (probably). Retinal activity in identification
Seredinskii,.A. V. (probably). Retinal activity_ in identification
Tsukkerman, 'I. I. (probably). Retinal ,activity in identification
Institute of Surgery inieni Vishnevskil: learning syStems, for recogni-
tion (location of images in the .brain); pattern-recognition techniques
for rapid assessment of burns
Braynes, S. N., Head of Neurophyslological Laboratory. Algorithms
of conditioned reflex development; .neurocybernetics; self-organiz-
ing systems
Academy of Pedagogical Sciences, RSF:SR
Institute of Psychology:
Leont'ev, A. N. Information processes in man
Moscow State Pedagogical Institute:
Gashcheriko, N. M. Recognition of meaningful sentences
Scientific Research Institute of Defectology:
Muratov, R. S. Reading devices
Academy of Sciences, USSR
Computer Center:
Kozhukin, G. I. Self-teaching machines
Institute of Automation and Telernechanics: learning systems for rec-
ognition (machine to discern handwritten numerical figures)
Ayzerman, M. A. Learning systems for recognition,
Bashklrov, 0. A. ,Learning systems for recognition
Braverman,. E. M. Learning systems for recognition: "compactness
hypothesis"
Feldbaurn, A. L Machine intelligence
Muchnlk, I B. Learning, systems for recognition
Shtirman, Ye. V. Modeling instruction process using psychological
learning theory
Approved for Release: 2018/03/28 C06731721
pproved for Release: 2018/03/28 C06731721
Institute of Biophysics: mathematical modeling of recognition proc-
esses; learning systems for recognition (identification of images by
means of indices)
Bongard, M. M. Mathematical modeling of recognition processes;
learning systems for recognition
Maksimov, V. V. Learning systems for recognition
Petrov, A. P. Learning systems for recognition
Smirnov, M. S. Learning systems for recognition
Vayntsvayg, M. N. Learning systems for recognition
Zenkin, G. M. Learning systems for recognition
Institute of Philosophy:
Novik, I. B. Modeling information processes
Institute of Physiology imeni Pavlov:
Anoishln, P. K. At Cybernetics Laboratory. Physiology and
cybernetics
Laboratory of Electromodeling: site of Scientific Technical Conference
on Cybernetics (1957)
Avrukh, M. L editor of a VINITI publication. Reading devices
Outenmakher. L. I. Director of Laboratory of Electromodeling.
Electrical modeling of thought processes; automating informa-
tion input
Kholsheva, A. F. Reading devices
Stretsiura, 0. G. Reading devices
Mathematics Institute and Computer Center (Novosibirsk): self-teach-
ing machines
Kozhuicin, 0. I. Self-teaching machines
Mathematics Institute irneni Steklov:
Kolmogorov, A. N. Parallel-operating automata; modeling think-
ing beings
Lyapunov, A. A editor, Problemy Kibernetild. General cybernetics
Lyubimskil, E. Z. Reading devices
Ofxnan, Yuri!. Parallel-operating automata
Mathematics Institute imeni Steklov, Leningrad Department:
Varshayskil, V. I. Minimum description of images required for arti-
ficial recognition; pattern recognition with learning
Party committee of the Presidium: Co-sponsored conference on "Phil-
osophical Problems of Cybernetics"
Scientific Council on Cybernetics: general coordination of cybernetics
research work; sponsored seminar on "Reading Devices"; co sponsored
conference on "Philosophical Problems of Cybernetics"
Parin, V, Chairman of Section on "Cybernetics and Living Nature"
(Bionics)
Prokhorov, A.
Scientific Council on "Philosophical Problems of Natural Sciences": co-
sponsored conference on "Philosophical Problems of Cybernetics''
Academy of Sciences, Latvian SSR
Institute of Electronics and Computer Technology:
Dambitis, Ya. Ya. (probably). Self-organizing systems
Institute of Physics:
Shneps, M. A Self-organizing systems
for Release: 2018/03/28 C06731721
, �
Approved for Release: 2018/03/28 C06731721
Academy of Sciences, Ukraine SSR
Computer Center:
Glushkov, V. M. Machine intelligence; self-teaching/self-organizing
systems; pattern recognition with learning; reading devices; rec-
ognition of meaningful sentences
Kovalevskii, V. A. Pattern recognition with learning; reading de-
vices; automation of information input
Stognii, A. A. Recognition of meaningful sentences
Cybernetics Council: sponsored seminars on "Automation of Thinking
Processes" and blocybernetics
Glushkov, V. M., Chairman of Cybernetics Council. Machine intelli-
gence; self-teaching/self-organizing systems; pattern recognition
with learning; reading devices; recognition of meaningful sen-
tences
Stognii, A. A., Scientific Secretary of Cybernetics Council. Recogni-
tion of meaningful sentences
Institute of Cybernetics:
Gluskov, V. M. Director. Machine intelligence; self-teaching/self-
organizing systems; pattern recognition with learning; reading
devices; recognition of meaningful sentences
Institute of Electrical Engineering:
Ivaktmenko, A. G. Control theory in artificial intelligence; percep-
tron-type device
Mathematics Institute:
Amosov, N. M., Leader of Seminar on Blocybernetics
Kukhtenko, A. I. Self-organizing (control) systems
Other Institutes and Personnel Associated with Artificial-Intelligence
First Moscow Medical Institute, Department of Physiology:
Anokhin. P. K., Head of Department of Physiology. Physiology and
cybernetics
Kazan Aviation Institute:
Borshche, V. B.. published in Trudy Kazan Aviatsionnyi Institut
(Works of the Kazan Aviation Institute). Machine intelligence
Il'in V. V., published in Trudy Kazan Aviatszonnyi Institut. Ma-
chine intelligence
Rokhlin, F. Z., published in Trudy Kazan Azdatstonnyi Institut. Ma-
chine intelligence
Kiev Computer Center:
Kondratov, A., associated with work at Kiev Computer Center; writer
on artificial intelligence
Kiev Institute of Civil Air Fleet imeni Voroshilov:
Kukhtenko, A. I. Self-organizing (control) systems
Laboratory (now Institute) for Systems for Transmission of Information:
Blokh, E. L. Minimum description of images required for artificial
----recognition; "compactness hypothesis"
Garmash, V. A. Quasi-topological approach to recognition; reading
devices
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Kharkevich, A. A. Minimum description of images required for
artificial recognition
Kirillov, N. Ye. Automatic discrimination of speech sounds
Pereverzev-Orlov, V. S. (probably). Quasi-topological approach to
recognition; reading devices
Tsirlin, V. M. Quasi-topological approach to recognition; reading
de vices
Latvian State University, Computer Center:
Aft', E. I. Self-teaching machines
Leningrad State University, Experimental Laboratory. of Machine Trans-
lation:
'Andreev, N. D. Reading devices
Military Air Engineering Academy imeni Zhukovsldy:
Chinayev, P. I Self-teaching/self-organizing systems
Moscow Power Engineering Institute: brain modeling
Kushelev, Yu. N.. Engineer. Neurocybernetics
Krug, 0. K. Self-teaching machines
Letskii, E. L. Self-teaching machines; neurocybernetics
Svechinskil, V. B., student. Neurocybernetics, modeling thought
processes
Moscow State University: site of continuing Cybernetics Seminar
Leont'ev, A. N. Information processing In man
Lyapunov, A. A.. coordinator of Cybernetics Seminars. General cy-
bernetics
Moscow State University, Department of Higher Nervous Activity:
Chichvarina, N. A. Algorithms of conditioned reflex development;
neurocybernetics
Napalicov, A. V. Algorithms of conditioned reflex development;
neurocybernetics; self-organizing systems
Semenova, T. P. Algorithms of conditioned reflex development;
neurocybernetics.
Shtirman, Ye. V. Modeling instruction processes using psychologi-
cal learning theory
Sokolov, Ye. N. modeling perception
Turov, A. F. Algorithms of conditioned reflex development; neuro-
cybernetics
Voloshinova, Ye., V. Algorithms of conditioned reflex development;
neurocybernetics; modeling instruction processes] using psycho-
logical learning theory
Order of Lenin Institute of Power:
Fercibrium, A. I., Faculty of Automation and Computer Technology.
Machine Intelligence
University imeni Palacki, Olomuc, Czechoslovakia, Chair of Psychology:
Golas, E. Generalization in pattern recognition and learning
Osladilova, D. Generalization in pattern recognition and learning
Valousek, C. Generalization In pattern recognition and learning
Zapbrozhe Oblast Psychiatric Hospital:
Gasul', Ya. R. Modeling thought processes
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721-
Additional Personnel Associated with Artificial-Intelligence
Biryukov, B. V., contributor to Problem Kirbernetiki
Blinkov, S. M. Structure of the brain
Fain, V. S. Minimum description of images required for artificial
recognition; learning systems for recognition; automating infor-
mation input
Fatkin, L. V., contributor to Voprosy Psikhologn (Questions of Psy-
chology). Philosophical problems of cybernetics; automatic dis-
crimination of speech sounds
Gutchin, Izrail', contributor to Zarya Vostoka (Dawn of the East),
Tbilisi, Georgian SSR
Kamynin, S. S. ,Reading devices
Krisilov, A. D. Reading devices
Kiibilyus, I., Prof., contributor to Kommunist (1111'nyus). General
cybernetics
Mayzel', N. I., contributor to Voprosy� Psikhologgi (Questions of
Psychology); philosophical problems of cybernetics
Mitulinskil, Yu. T. Minimum, description of images required for
artificial recognition
Petrenko, A. I., contributor to lzvestiya Vuz (News of Higher Edu-
cational Institutions). Reading devices
Pollakov, V. G., contributor to lzvestiya AN SSSR (News of the
Academy of Sciences, USSR). Reading devices
Rozhanskaya, E. V. Mathematical theory of intelligibility .(recogni-
tion)
Rybak, V. I. (probably Computer Center AS, UkSSR).. Pattern rec-
ognition with learning
Saenko, G. I., editor of VINiTipublication. Reading devices
Saparina, Ye. Brain modeling
Semenova, T. N. Pattern recognition with learning
Sindlievich, L. M., editor of VINITI,publication. Reading devices
Sokolovskii, V. A. Minimum description of images required for
- artificial recognition
Svechnikov, S. V.. contributor to lzvestiya VUZ (News of Higher Ed-
ucatlonal Institutions). Reading devices
Tarakanov, V. V. Perceptual activity in man
Tiuichtin, V. S. Theory of Images (and Perception)
Tsemel', G. I. Automatic discrimination of speech sounds
Visil'ev, A. M.. editor of VINITI publication. Reading devices
Vitushkin, A. G. Reading devices
Wang, Chih-ch'ing. Perceptual activity In man
Yeliseyev, V. K. Modeling recognition process
Zinchenko, V. P. Perceptual activity In man
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
REFERENCES
1. Lyapunov, A. A. "Certain General Prob-
lems of Cybernetics," Problemy Kiberne-
tiki (Problems of Cybernetics), v 1, Mos-
cow, 1958, p 5 (Pergamon Press, London,
1960)
2. PaTin, V. and A. Prokhorov. "Uncovering
the Secrets of Living Nature," Moscow,
Pravda (Truth), 24 Mar 63
3. Biryukov, B. V. "Conference on Philo-
sophical Problems of Cybernetics," Pro-
blemy Kibernetiki (Problems of Cyber-
netics), v 9, Moscow, 1963, p 349 (JF'RS:
21,448, 14 Oct 63, p 611)
4. Papers related to artificial-intelligence
presented at the Conference on Philo-
sophical Problems of Cybernetics as listed
in 3 above:
Leont'ev, A. N. "On Some Particulars of
the Processing of Information by Man"
Kolmogorov, A. N. "Life and Thinking
from the Point of View of Cybernetics"
Lyapunov, A. A. "On Control Systems in
Living Nature and General Understand-
ing of Living Processes"
Glushkov, V. M. "Thinking and Cyber-
netics'! _
Novik, I. B. "On the Nature of Infor-
mation and the Peculiarities of Cyber-
netic Modeling"
Fel'dbaum, A. I. "Instruction Processes
for People and for Automata"
5. Mayzel', N. I. and L. V. Fatkin. "Confer-
ence on Philosophical Problems of Cyber-
netics," Voprosy Psikhologii (Questions of
Psychology), Moscow, no 5, 1962 (JPRS:
17,574, 12 Feb 63)
6. Il'in, V. V., Borshche, V. B., and Rokhlin,
F. Z. "Can a Machine Think?", Trudy
Kazansk Aviatsionnogo Instituta (Works
of the Kazan Aviation Institute), Issue
65, 1961 ( Ref erativnyi Z hurnal�M ate-
-matika) (Reference-Journal�Mathema-
tics), no 2, Feb 63, Abst 2 V213)
7. Gasul', Ya. R. "On the Possibility of Ob-
taining a Model for Thought Functions,"
Kiev, Fiziologicheskiy Zhurnal (Physio-
logical Journal), v 8, no 6, Nov-Dec 62
8. Glushkov, V. M. "Thought and Cyber-
netics," Voprosy Fzlosofii (Questions of
Philosophy), Moscow, v 17, no 1, 1963, p
36-48 (JPRS: 18,302, 22 Mar 63)
9. Kondratov, A. "Automata, Thinking and
Life," Znaniye-Szla (Knowledge is Pow-
er), Moscow, no 6, 1962 (JPRS: 15,851, 23
Oct 62)
10. Kubilyus, I. "Cybernetics and Life,"
Virnyus, Kommunist (Communist), no 4,
Apr 63
11. Minsky, Marvin. "Steps Toward Artificial-
Intelligence," Proceedings of the IRE, v
49, no 1, Jan 61
12. � . "A Selected Descriptor-Indexed
Bibliography to the Literature on Artifi-
cial-Intelligence," IRE Transactions on
Human Factors in Electronics, 1961
13. Gruenberger, Fred. "Benchmarks in Ar-
tificial-Intelligence," Datamation, v 8, no
10, Oct 62, p 3-35
14. Berg, A. "Cybernetics and Its Domain,"
Vechernyaya Moskva, 11 Nov 63, p 2
� Keldysh, M. V. "The Development of Cy-
bernetics is the Concern of Scientists
from Various Specialties," Technical Cy-
bernetics, no 3, 1963 (JPRS: 21,100, 16
Sep 63, p 4)
16. Lincoln Laboratory Library. 8th refer-
ence bibliography, Artificial-Intelli-
gence�Soviet Bloc, Morris D. Friedman,
compiler, 27 Dec 61
17. Rozhanskaya, E. V. "On the Question of
the Mathematical Foundation of the
Theory of Intelligibility," Trudy, Komitet
po Akustiki (Works, Committee on Acous-
tics); no 7, 1953
15
11
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
18. Kamynin, S. S. and Lyubimskiy, E. Z.
"Principles of Reading Symbols Using
Automatic Machines," at Scientific-Tech-
nical Conference on Cybernetics (1957)
as reported in Pro blemy Kibernetiki
(Problems of Cybernetics), v 1, 1958, p
266 (Pergamon Press, London, 1960)
19. Kharkevich, A. A. "Pattern Recogni-
tion," Radiotekhnika (Radio Engineer-
ing), v 14, no 5, 1959
20. Sokolov, E. N. "A Probability Model of
Perception," Voprosy Psikhologii (Ques-
tions of Psychology), v 6, no 2, 1960
21. Tiulchtin, V. S. "K Probleme Obraza"
(On the Problem of Image), Voprosy Filo-
sofii (Questions of Philosophy), no 6, 1959
22. Blokh, E. L. "Toward the Question of
Minimum Description," Radiotekhnika
(Radio Engineering), v 15, no 2, Feb 60
23. Fayn, V. S. "On the Quantity of Coordi-
nate Descriptions of Images in Sys-
tems for Visible Pattern Recognition," Iz-
vestiya AN SSSR, OTN, Energetika i Avto-
matika (News of the Academy of Sci-
ences, USSR, Department of Technical
Sciences, Power Engineering and Auto-
mation), no 2, 1960
24. "On an Abbreviated Inscription
of Absolute Descriptions of Images," lz-
vestiya AN SSSR, OTN, Energetika i Avto-
matika, no 1, 1961
25. Varsha.vskii, V. I. and Sokolovskii, V. A.
"On the Problem of Recognizing Configu-
rations," Kibernetilca i Etektronno-Vy-
chisliternaya Tekhnika po Mat erialam
XVI Nauchno-Tekhnicheskoi Konferen-
tsit (of Nauchno-Tekhnicheskoye Ob-
shch estvo Radiotekhniki i Elektrosvyazi
im A. S. Popova) (Cybernetics and Elec-
tronic-Computing Technology�Materials
of the 16th Scientific-Technical Confer-
ence�of the Popov Society), Moscow/
Leningrad, Gosenergoizdat, 1962, p 12
26. Mitulinskii, Yu. T. "Recognition of Digi-
tal Symbols," Voprosy Vychisliternoy
Tekhniki (Questions of Computer Engi-
neering), Kiev, Gostekhizdat UkSSR, 1961
(JPRS: 16,490, 3 Dec 62, p 21)
27. Kharkevich, A. A "Selection of Criteria
in Mechanical Recognition," lzvestiya
AN SSSR, OTN, Tekhnicheskaya. Kiber-
netiki (News of the Academy of Sciences,
USSR, Department of Technical Sciences,
Technical Cybernetics), no 2, Mar/Apr
63 (JPRS: 19,966, 1 Jul 63, p 1)
28. Ayzerman, M. A. "Experiments on
Teaching Machines to Recognise Visual
Images," Biological Aspects of Cyber-
netics�Collection of Works, Depirtment
of Biological Sciences, Scientific council
on the Complex Problem "Cybernetics,"
Academy of Sciences USSR Publishing
House, Moscow, 1962 (JPRS: 19,637, 11
Jun 63, p 242)
Blokh, E. L. "Some Approaches to a
Practical Solution of the Problem of Rec-
ognition," Izvestiya . AN SSSR, OTN,
Tekhnicheskaya Kibernetiki (News of the
Academy of Sciences, USSR, Department
of Technical Sciences, Technical Cyber-
netics), no 2, Mar/Apr 63 (JPRS: 19,966,
1 Jul 63, p 14)
30. Zinchenko, V. P., Wang Chih-ch'ing and
V. V. Tarakanov. "The Establishment
and Development of Perceptual Activity,"
Voprosy Psikhologii (Questions of Psy-
chology), Moscow, no 3, 1962 (JPRS:
16,404, 29 Nov 62, p 1)
31. Glezer, V. D, A. A. Nevskaya, A. V.
Seredinskli and I. I. Tsukkerman. "On
the Identification of Objects in the Visual
System," Biologicheskiye Aspekty Kiber-
netiki�Sbornik Rabot (Biological Aspects
of Cybernetics�Collection of Works), De-
partment of Biological Sciences, Scien-
tific Council on the Complex Problem
"Cybernetics," Academy of Sciences USSR
Publishing House, Moscow, 1962 (JPRS:
19,637, 11 Jun 63, p 225)
32. Avrukh, M. L., A. F. Kholsheva, and G. G.
Stretsiura. "Reading Device for an In-
formation Machine," Scientific-Technical
Conference on Cybernetics (1957) (see 18
above)
33. Andreev, N. D. "Principles of Construc-
tion of Electronic Reading Devices," at
All-Union Machine Translation Confer-
29.
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
Lion of Criteria
ion," lzvestiya
heskaya Kiber-
my of Sciences,
3nical Sciences,
lo 2, Mar/Apr
p 1)
cperiments on
:cognize Visual
!cts of Cyber-
ts, Department
ientilic Council
"Cybernetics,"
;SR Publishing
'RS: 19,637, 11
)proaches to a
?roblem of Rec-
SSSR, OTN,
ici (News of the
12., Department
achnical Cyber-
(JPRS: 19,966,
;13113-ch'ing and
Establishment
ptual Activity,"
estions of Psy-
, 1962 (JPRS:
evskaya, A. V.
kkerman. "On
As in the Visual
Aspekty Kiber-
*logical Aspects
of Works), De-
sciences, Scien-
dnplex Problem
f Sciences USSR
w, 1962 (JPRS:
theva, and G. G.
!vice for an In-
intific-Technical
s (1957) (see 18
les of Construe-
ng Devices," at
islation Confer-
ence (1958), as reported in Problemy
Kibernetiki (Problems of Cybernetics),
v 2, 1959, p 307
34. Muratov, R. S. "Device for Reading of
Ordinary Typographic Text by the
Blind," All-Union Machine Translation
Conference (1958) (see 33 above, p 308)
35. Kharkevich, A. A. "On Principles of
Constructing Reading Machines," Radio-
tekhnika (Radio Engineering) , v 15, no
2, Feb 60
36. Fayn, V. S. "On Principles of Construct-
ing Machines for Recognition of Pat-
terns," Radiotekhnika (Radio Engineer-
ing), v 15, no 3, Mar 60
37. Petrenko, A. I. and Svechnikov, S. V.
"Main Trends in the Development of
Reading Automata," lzvestiya Vysshikh
Uchebnykh Zavedeniy, MVySSO, Radio-
tekhnika (News of Higher Educational
Institutes, Ministry of Higher and Spe-
cialized Secondary Education, Radio En-
gineering), v 4, no 3, 1961 (JPRS: 11,524)
38. Fayn, V. S. "On Automation of the In-
put of Certain Kinds of Data into Com-
puters," Avtonuttika i Telemekhanika
(Automation and Telemecha.nics), v 22,
no 4, 1961
39. Gutenmakher, L. I. "Elektronnye Infor-
matsionno-Logicheskie Mashiny" (Elec-
tronic Information-Logical Machines),
2nd Ed., Moscow Academy of Sciences
Press, 1962, p 36-49
40. Kovalevskii, V. A. "Construction of Read-
ing Automata for Electric Digital Ma-
chines," at Kiev Cybernetics Section, as
reported in Problemy Kibernetiki (Prob-
lems of Cybernetics), v 6, 1961
41. Biological Aspects of Cybernetics�Collec-
tion of Works (Biologicheskiy Aspekty
Kibernetiki�Sbornik Rabot), Department
of Biological Sciences, Scientific Council
on the Complex Problem "Cybernetics,"
Academy of Sciences USSR Publishing
House, Moscow, 1962 (JPRS: 19,637, p
331-334)
42. Pereverzev-Orlov, V. S. and Poliakov, V. G.
"On the Construction of Reading Ma-
chines," lzvestiya AN SSSR, OTN, Ener-
getika i Avtomatika (News of the Acad-
emy of Sciences, USSR, Department of
Technical Sciences, Power Engineering
and Automation), no 3, 1961, p 110-112
43. . "Universal Automat for Reading
Printed Text," in Chitayushchiye Ustroi-
stva (Reading Devices), Moscow, 1962, p
73-82 (JPRS: 18,119, 13 Mar 63, p 169-
190)
44: Garrnash, V. A. "Seminars on Reading
Devices," Vestnik AN SSSR (Herald of
the Academy of Sciences, USSR), v 30,
no 9, 1960
45. Bongard, M. M. "Recognition Modeling
on a Computer," at Symposium on Prin-
ciples of Design of Self-learning Systems
(1961) as reported in Problemy Kiber-
netiki (Problems of Cybernetics), v 7,
Moscow, 1962, p 231-32
46. . "Modeling Recognition Proc-
esses," at Cybernetics Seminar, Moscow
State University, 1961, as reported in
Problemy Kibernetiki (Problems of Cy-
bernetics), no 7, p 233
47. . "Modeling the Process of In-
structing for Recognition on a Universal
Computer," Biologicheskii Aspekty Ki-
bernetiki�Sbornik Rabot (Biological As-
pects of Cybernetics�Collection of Works)
(see 41 above) (JPRS: 19,637, p 261)
48. Glushkov, V. M. "Designing Principles
of a Universal Reading Automaton," Av-
tomatYka (Automation), no 1, Kiev, 1962,
p 55-64 (JPRS: 15,303, 17 Sep 62)
49. "A Machine That Reads," Pravda Ukrainy
(Ukraine Pravda), 12 Jan 62 (FrD-TT-
62-1205)
50. Yeliseyev, V. K. and Kovalevskiy, V. A.
"Study of an Algorithm for the Recogni-
tion of Machine-Written Symbols,"
Zhurnai Vychisliternoy Mat ematiki i
Matematicheskoi, Fiziki (Journal of Com-
puter Mathematics and Mathematical
Psysics), v 2, no 5, Sep/Oct 62 (JPRS:
16,343, 26 Nov 62)
51. Academy of Sciences, USSR, Institute of
Scientific Information, Avrukh, M.
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
Vasirev, A. M., Saenko, G. I., and Sinde-
levich, L. M. (eds.), Reading Devices, Mos-
cow, 1962, 188 pp (JPRS: 18,119)
52. Braverman, E. M. "Some Problems of
Design of Machines Which Classify Ob-
jects According to� Criteria Previously
Unknown," Avtomatika i Telemekhanzka
(Automation and Telemechanics), Mos-
cow, no 10, 1960, p 1375-1386 (JPRS:
9028, 27 Apr 61)
53. Glushkov, V. M., Kovalevskiy, V. A., and
Ryl:�ak, V. I. "On One Algorithm for
Teaching the Recognition of Patterns,"
at Symposium on Principles of Design of
Self-learning Systems (1961) (see 75 be-
low)
54. Semenova, T. N. and Varshayskii, V. I.
"A Learning Program for Recognition of
Configurations," at Symposium on Prin-
.ciples of Design of Self-learning Systems
(1961) (see 75 below)
55. Gutchin, Izrail'. "Animals and Ma-
chines�Cybernetics Performance," Zarya
Vostoka (Dawn of the East), Tbilisi, 28
Jul 62, p 4
56. Ayzerman, M. A. "Experiments with
Teaching a Machine to Recognize Pat-
terns Without the Introduction of Their
Properties [Descriptions] into the Pro-
gram of the Machines," Cybernetics Sem-
inar at Moscow State University, 1962, as
reported in Problemy Kibernetiki (Prob-
lems of Cybernetics), v 9, Moscow, 1963,
p345
57. "They Teach a Machine to Rec-
ognize Visual Images," Nauka i Zhizn
(Science and Life) ,,no 12, Dec 62, p 34-39
(JPRS: 18,126, 14 Mar 63)
58. Varshayskii, V. I. and Semenova, T. N.
"Learning Program for Recognition or
Configurations," Kibernetika i Elektron-
no-Vychisliternaya Tekhnika (Cybernet-
ics and Electronic-Computing Technol-
ogy) (see 25 above)
59. . "Teaching the Identification of
Configurations Formed of the Simplest
Signs," at Symposium on Principles of
Design of Self-learning Systems (1961)
(see 75 below)
�
60. Glushkov, V. M., Kovalevskii, V. A., and
Rybak, V. I. "Algorithm of a Learning
Machine for Recognition of the Simplest
Geometric Figures," at Symposium on
Principles of Design of Self-learning Sys-
tems (1961) (see 75 below)
61. Glushkov, V. M., Grishchenko, N. M., and
Stognii, A. A. "Algorithm for the Recog-
� nition of Meaningful Sentences," at Sym-
posium on Principles of Design of Self-
learning Systems (1961) (see 75 below)
62. Tsemel', G. I. "Automatic Discrimina-
tion of Invariant Criteria of Occlusive
(Plosive) Sounds by Means of Clipped
Speech Signals;" lzvestiya AN SSSR,
07'N, Energetika iAvtomatika (News of
the Academy of Sciences, USSR, Depart-
ment of Technical Sciences, Power Engi-
neering and Automation), no 5, 1960
63. Kirillov, N. Ye. and Fatkin, L. V. "Ex-
periments in the Recognition of Speech
Sounds by Automatic Machines," Vo-
prosy Psikhologii (Questions of Psychol-
ogy), Moscow, no 3, May/Jun 62, p 45-55
(JPRS: 15,745, 17 Oct 62)
64. Control, London, v 7, no 65, Nov 63, p 262
65. Kozhukin, G. I. "A Programmed Model
of a Self-teaching Machine," All-Union
Conference on Computing Mathematics
and Computer Technology (1960), as re-
ported in Problemy Kibernetiki (Prob-
lems of Cybernetics), v 5, Moscow, 1961
66. Arin', E. I. "On Self-Teaching Machines,"
Cybernetics Seminar at Moscow State
University, 1960, as reported in Problerny
Kibernetiki (Problems of Cybernetics),
Moscow, no 5, 1961
67. Glushkov, V. M. "Principles of Con-
structing Self-teaching Systems," Kiev
Section on Cybernetics, as reported in
Problemy Kibernetiki (Problems of Cy-
bernetics), Moscow, no 6, 1961, p 301
68. Rosenblatt, F. "Perceptual Generaliza-
tion Over Transformation Groups" in
Self-Organizing Systems, Yovits, M. C.
and Cameron, S. (eds.), N. Y., Pergamon
Press, 1960, p 63
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
�
kii, V. A., ana
of a Learning
1 the Simplest
;ymposium on
f-learning Sys-
')
iko, N. M., and
for the Recog-
awes," at Sym-
)esign of Self-
(see 75 below)
lc Discrimina-
a of Occlusive
ins of Clipped
ya AN SSSR,
atika (News of
USSR, Depart-
Power Engi-
, no 5, 1960
in, L. V. "Ex-
tion of Speech
dachines," Vo-
)ns of Psychol-
Jun 62, p 45-55
5, Nov 63, p 262
;rammed Model
me," All-Union
g Mathematics
y (1960), as re-
ernetiki (Prob-
1. Moscow, 1961
ring Machines,"
Moscow State
ted in Problemy
4 Cybernetics),
eciples of Con-
Systems," Kiev
as reported in
Voblems of Cy-
^ 1961, p 301
Generaliza-
on Groups" in
� Yovits, M. C.
N. Y., Pergamon
69. . "A Comparison of Several Per-
ceptron Models" in Self-Organizing Sys-
tems 1962, Yovits, M. C., Jacobi, G. T.,
and Goldstein, G. D. (eds.), Washington,
Spartan Books, 1962, p 463
70. Ivalchnenko, A. G. "Application of the
Invariance Theory and Combined Control
in the Synthesis and Analysis of 'Learn-
ing' Systems," Avtomatyka (Automation),
Kiev, no 5, 1961, p 3-12 (FTD-TT-62-75,
30 Mar 62)
71. Voloshinova, Ye. V. and Shtil'man, Ye. V.
"On the Modeling of the Processes of In-
struction in Automatic Systems," in
Sbornik: Avtomaticheskoye Regulirova-
niye i Upravleniye (Collection: Automatic
Regulation and Control) Moscow, Acad-
emy of Sciences, USSR, 1962 (from Refe-
rativnyi Zhurnal�Matenzatilca (Refer-
ence Journal�Mathematics), no 2, Feb
63, Abst 2 V523)
72. Yovits, M. C. Preface to Proceedings of
the First Conference (1960), in Self-Or-
ganizing Systems 1962, op. cit., p viii
73. Braynes, S. N. and Napalkov, A. V.
"Nekotorye voprosy teorii samoorganizyu-
shchikhsya sistem" (Some Questions of
the Theory of Self-Organizing Systems),
Voprosy Filosofii (Questions of Philos-
ophy), Moscow, no 6, 1959, p 148-154
74. Dambitis, Ya. Ya. and Shneps, M. A.
"On Self-Organizing Systems," All-Union
Conference on Computing Mathematics
and Computer Technology (1960) (see 65
above)
75. "Principles of Design of Self-learning
Systems," Kiev, Gostekhizdat, 1962
(JPRS-: 18,181, 18 Mar 63) and reported
in Problemy Kibernetiki, v 7, Moscow,
1962, p 231-232
76. Chinayev, P. I. "Self-learning�One of
the Basic Means of Development of Self-
regulating Systems," at Symposium on
Principles of Design of Self-learning Sys-
tems �(see 75 above)
77. Kukhtenko, A. I. "On Self-Organizing
Control Systems," at Symposium on
Principles of Design of Self-learning Sys-
tems (1961) (see 75 aboye)
78. -6el1-Organiz1ng 6ynems, � Yovits, M. U.
and Cameron, S. (eds.), op. cit.
79. Anokhin, P. K. "Fiziologiya i Kiber-
netiki" (Physiology and Cybernetics),
Voprosy Filosofii (Questions of Philos-
ophy), no 6, 1955
80. Gutenmakher, L. I. "Elektricheskoe
modelirovanie nekotorykh protsessov um-
stvennogo truda s pomoshch'yu informa-
tsionnykh mashin s bol'shoi vnytrennei
pamyat'yu" (Electrical modeling of some
mental work processes using information
machines with large internal memory),
Scientific-Technical Conference on Cy-
bernetics (1957), as reported in Problemy
Kibernetiki (Problems of Cybernetics),
v 1, 1958,.p 266 (Pergamon Press, London,
1960)
81. Blinkov, S. M. "On the Structure of the
Brain," Cybernetics Seminar at Moscow
State University, 1960, as reported in
Problemy Kibernetiki, v 5, Moscow, 1961
82. Braynes, S. N., Napalkov, A. V., and
Svechinskiy, V. B. "Scientific Notes
(Problems of Neuro-cybernetics)," Mos-
cow, 1959, 109 pp (JPRS: 5880, 18 Nov
60)
83. Napalkov, A. V. "Nekotorye Printsipy
Raboty Golovnogo Mozga" (Some Prin-
ciples of Operation of the Brain), Pro-
blemy Kibernetiki (Problems of Cyber-
netics), v 4, Moscow, 1960, p 183
84. . "Information Processing by the
Brain," in Biologicheskiye Aspekty Kiber-
netiki (Biological Aspects of' Cybernetics) ,
Academy of Sciences USSR Publishing
House, Moscow, 1962 (JPRS: 19,637, 11
Jun 63, p 174)
85. Golas, E. "Study of the Conditions of
Generalization," Voprosy Psikhologii
(Problems of Psychology), Moscow, no
3, 1962 (JPRS: 16,404, 29 Nov 62, p 55)
86. Stognii, A. A. "On Coordination of Sci-
entific-Research Work on Cybernetics in
the Ukraine," Problemy Kibernetiki
(Problems of Cybernetics), v 9, Moscow,
1963, p 342 (JPRS: 21,448, 14 Oct 63, p
598-600)
1-3
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
� Scientific
Intelligence
Report
FOR -OFFIGAL�USE�GM�Y-- N? 135
rtificial-Intelli � ence Research ,
In the US$;,
aq
PUBLICATIONS FILE COPY
DO NOT, REMOVE
OSI�SR/64-37
8 September 1964
Office of Scientific Intelligence
Approved for for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
WARNING
Laws relating to copyright, libel, slander, and
communications require that the dissemination of
this publication be limited to
/ Exception can be granted only by the is-
suing agency, and users are warned that noncom-
pliance may subject violators to personal liability.
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Scientific Intelligence Report
ARTIFICIAL -INTELLIGENCE RESEARCH
IN THE USSR
OSI�SR/64-37
8 September 1964
CENTRAL INTELLIGENCE AGENCY
OFFICE OF SCIENTIFIC INTELLIGENCE
FOR cTricIAL USE ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Project Officer
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL USE ONLY
PREFACE�
Artificial-intelligence is a popular term that was coined
in the United States during the 1950's to categorize research
studies aimed at simulation of intelligent or "thinking" be-
havior. This type of research seeks to analyze the factors in-
volved in the making, by humans, of specific types of decisions,
to specify and define these factors as decision procedures or
mathematical algorithms, and ultimately to fabricate hard-
ware which can concretely model decision procedures and
thereby assist human decision makers in making increasingly
complex decisions. Although current digital computers can as-
sist in the solving of some complex decision problems, artificial-
intelligence research has already discovered decision routines
which, while they can be modeled on a digital computer for
demonstration purposes, are more efficiently solved with other
types of hardware. In the future this will increasingly involve
some sort of analog or combined analog-digital equipment, pos-
sibly a replica of the structure of the human brain, more likely
a model abstracting essential elements of brain function on
principles not yet uncovered.
Artificial-intelligence research is necessarily interdiscipli-
nary in nature, involving such traditional areas as biology, phys-
iology, psychology and electrical and systems engineering, with
a strong under-laying of mathematics. As used in US literature,
the term "artificial intelligence" may appear to be a rather
flexible "tent," encompassing more or less whatever one desires
to place within it. Nevertheless, artificial-intelligence is a "far
out" field of scientific endeavor, the surface of which has been
merely scratched. Its potential for future accomplishments
can be only dimly seen and not evaluated at present. If it
succeeds in significantly optimizing decision making in such
complex areas as the economy or national strategic planning,
it will obviously make a strong contribution to a relatively mon-
olithic system, such as the Soviet one.
In the USSR, research on approaches to the fabrication of
problem-solving or decision-making machines (that is, artificial-
intelligence research) is conducted under the general category
of cybernetics. This report covers Soviet research on major
problems in this field, including pattern recognition by ma-
chines, machine learning, planning and induction in the prob-
lem-solving machine, and brain modeling. It does not cover
conventional digital computer solution of problems or on-line
computer control of processes.
The material in this report is based chiefly on information
from the Soviet open literature available as of 1 May 1964.
Additional information obtained through 1 July 1964 does not
materially affect the conclusions.
FOR orricIAL, UCE ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
MR OFFICIAL USE ONLY
CONTENTS
Page
PREFACE
PROBLEM
1
CONCLUSIONS � ''' � � �
1
SUMMARY
1
DISCUSSION
2
Introduction
2
Philosophical aspects . ......... .
3
Principal research problems
6
Search and pattern recognition
7
Learning, induction, and planning . ... .
9
Brain modeling .... . . . . . . . . ... . .
11
APPENDIX � Institutes and Scientists Associated with
Research of Artificial-Intelligence Signifi-
cance in the Soviet Bloc
15
REFERENCES
21
561/07
ron. OFFICIAL USE ONLY vii
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
OR OFFICIAL USE, ONLY,
ARTIFICIAL-INTELLIGENCE RESEARCH
IN THE USSR
PROBLEM.
To iiSisa artificial-intelligence research in -the USSR:
CONCLUSIONS
1. In the Soviet Union, substantial govern-
mental and Party support and encouragement
are being given to extensive studies on the
theory of artificial-intelligence. .Soviet sci-
entists of high caliber ,are conducting arti-
ficialLintelligence studies and are eiposing.
young students to the state of the art in
this interdisciplinary field; laying a sound
foundation for further advances.
2: The importance'which the Soviet regime
attaches to artificial-intelligence research is
attested by the unusual freedom of open .dis-
cussion allowed scientists working in the field;
as reflected in the published literature.
3. Soviet research - in the theoretical and
hardware aspects of artificial-intelligence is
now about as ,advanced as US work and can
be expected to continue at a ,rate equal to or
greater than that observed in the West. Sig-
nificant theoretical achievements within the
next 5 years ,are highly probable. When
theories are converted into designs, Soviet en-
gineers probably will be able to produce the
_equipment.
SUMMARY
After a relatively late start, Soviet research
on artificial intelligence related to the ulti-
Mate development of decision-making, ma-
chines-now is about on a par with similar US
work. The apparently greater rate Of -Soviet
progress compared with that of the West is
attributable to the magnitude. of official 'rec-
ognition and support of artificial-intelligence
research in the USSR. Soviet officials con-
sider the development of decision-making ma-
chines to be essential to the successful man-
agement of their increasingly complex eco-
nomic and social system, and are giving sub-
stantial support and Party encouragement to
artificial-intelligence studies.
FEP OPitICIA.i. USE" ONLY
1
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
OPTIC= 1:75z. ora.x
Soviet ideologists, and scientists are discuss-
ing, from a wide range of viewpoints, the
fundamental philosophical problems about
the nature of intelligence and of man which
are raised by attempts to model decision
making or -intelligent behavior. The stric-
tures of dialectical materialist ,dogma are not
Inhibiting research in this area. In view of
the high prestige of the Soviet scientists tak-
ing part in these discussions and the publica-
tion in the Soviet scientific literature of the
opinions expressed, artificial intelligence stud-
ies represent rare and significant examples
of intellectual freedom in the USSR
Current Soviet work in the major subfields
of artificial-intelligence research includes in-
vestigation of techniques for machine search
of problem solutions, pattern recognition, ma-
chine learning, planning and inductiori in
machine systems, and. brain modeling. Soviet
progress in research on machine techniques
for search, pattern recognition, and learning
compares favorably with US advances. Sci-
entists and engineers engaged in research. on
pattern-recognition in the Soviet Union have
made -original contributions to the theoreti-
cal basis for artificial pattern-recognition, sys-
tems and have,. fabricated , various types of
pattern-recognition devices.
Soviet 're.search on simulation of such as-
pects of the cognitive process as learning and
induction at present consists mainly of study
of self-teaching and self-organizing systems.
In these fields Soviet scientists and engineers
have created models based on neuro-physio-
logical conditioned-reflex approaches, on psy-
chological learning theory, and on automatic
control system theory.
Soviet study of the exceedingly complex
problems involved in incorporating, techniques
of planning and inductiOn in artificial prob-
lem-solvers is at, a very early stage of devel-
opment. As-yet, little progress has been made
in these fields in the West.
Research on the modeling of brain processes
is receiving much emphasis in the USSR.
Since 1955, laboratory experimentation on
models that simulate neural processes has in-
Creased, and outstanding neuro-physiologists
and psychologists have begun to work closely
with mathematicians, computer specialists,
and electronics engineers on cybernetic stud-
ies of, brain-modeling problems.
Seminars are regularly offered, to expose
students to the state of the art in the fields,
that are especially significant to the future
of artificial intelligence research in the USSR.
These seminars are conducted by the leading
researchers in artificial-lintelligence and are
thereby laying a _sound foundation for future
'advances in a new and multidisciplinary' field.
DISCUSSION
INTRODUCTION'
Since the mid-1950's, the Soviet govern-
ment has increasingly emphasized the impor-
tance of a cybernetics research program for
achieving national and international goals..!
A large part of the; theoretical research aspect
of this program has been focused on the, de-
velopment of decision-making machine&
Realization of such machines wou1d assist the
Soviets in solving two types of fundamental
"AEC: 820/46, "The Meaning of Cybernetics
the USSR," Cybernetics in the USSR, 30 Mar 04
2
problems: (i) decision making on the basis of
incomplete data (sometimes referred to, a&de-
cision making under conditions of' ,uncer-
tainty), and (fl) decision making in the: pres-
ence of complete but overwhelming quantities
of data. Problems of the latter type are by
definition beyond the capabilities of human
decision makers.' Problems of the first type,
while inherently characteristic of human
thinking activity, are rapidly extending
beyond the:bounds of possible human, solution
in the context of ,economic management and
state:administration.
FOR OFFICIAL USE ONLY
C
(b)(3)
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Decision-making machines are, in, essence,
models of one Variety or another of the human, .
problem-solving proCess. Models of the hu-
man thinking or problem-Solving process are
artificial simulations Of the information flow,
or information processing, procedures of the
human problem solver. The model May be
mathematical; in which case it may be a pro
gram for a Universal. (digital) computing ma-
chine,- or it may be a piece of hardware, such
.as an electronic circuitry analog. (Many de-
cision procedures �already elaborated- by arti-
ficial-intelligence research are not efficiently
modeled by the single track, sequential:meth-
od of operation of conventional digital com-
puters. This would not be surprising in view
� of the apparent parallel operation of infor-
mation,processing in the human brain. Such
decision routines can, be modeled for demon-
stration purposes on a digital computer, but
analog models of some sort Will loom ,increas-
ingly impartantin;the'fabrication2of,artificial-.
decision Makers.) In any case, the essential
feature of such a model is that for a given
input Of information the Model shall produce
at least the Same output (Of information) as
the natural process being modeled. Re-
search on, such models is categorized in the
United States under the rubric "artificial=iii-
telligence."
,Soviet research On artifiCial-inteliigence has
mushroomed .since '1956 along with .other Sub-
problems of cyberneticS. After the establish-
Merit hi 1959 of a national program. for 'cy-
-bernetics, thedirection� of Soviet research' on
artificial-intelligence became one of the re-
sponsibilities of, the Scientific Problem:Coun-
cil on .Cybernetics of the Presidium, Academy
of �Sciences, USSR The Cybernetic Machines
Section of the Council appears to be the 'unit
which: assigns, monitors, and integrates- most
of this research work. on a national scale.
.,Actual research is conducted at the labora-
tories and institutes listed in the :appendix.
The. number- of facilities engaged in this
re-
search can be expected to grow as the-cyber-
netic approach continues to permeate other
areas of Soviet science and technology
cable� to communication and control 'problems
arising in productiori.industries, the economy,
,law, government administration, and military
activities.' -
Although much of the application-oriented
cybernetics' research is at present, directed to-
Ward the realization of more Sophisticated
conventional systems for automatic control
of Machines, plants, space research vehicles,
or even ecOnOmic units; a large part! of the
theoretical research is related, "directly or -in-
directly, to the development of "thinkinma-
ehines,',' or "thinking cybernatons," as Soviet
,scientists often refer. to them; Realization Of
such machines ' will have immediate �relevance
'to, information abstracting and retrieval; ma-
chine translation, and general problem solv-
ing, and eventual application to later-genera-
tiOn automatic control� systems. !.
A. A. LyaPunov, a leading 'Soviet mathema-
tician and editor of Problems or Cybernetics,,
has described the relationship of artificial-in-
telligence to :general Cybernetics, and the' na-
ture of: future contrOl_systems. He: suggests
that algOrithinization (mathematical model-
ing) of thinking processes and Of control pi*-
esses is one of the most ,important aspects of
cybernetic theory.1 Other members .of the
_Scientific, Council on Cybernetics, in explain-
Mg the importance of artificial-intelligence,
'state. that the basic, ,products - of radioelee-
tronics are various devices for automatic reg-
ulation, monitoring, control, and. ccirriinuilica-
bons. The brain, 'Which fulfills- these 'func-
tions in the living organism, Works much more
reliably and productively than any PreSent-
day ,radioeleatrOnic 'machine. . . . Some fit:
steps have already been taken in the direction
Of constructing practical automata analo-
gous to the *in." 2
�
� �
-PHILOSOPHICAL ASPECTS ;:�!;�
Reexamination of basic philOsophical atti-
tudes toward: the nature of the brain, of mind,
,and of man 'himself is taking -place wherever
artificiaUntelligence research is being 'con-
ducted. The Soviet Union is no exception'.
Philosophical theoretical questions bearing on
the simulation of intelligence, or intelligent
behavior; are being thoroughly .aired and in-
FOR OFFICIAL DOE ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFPICTAT TTSE ONLY
vestigated in Soviet scientific circles, outside
as well as within the context of the funda-
mentals of dialectical materialism.2-8
A highly significant discussion, took place
during June 1962 at a conference in MOSCOW
on "The Philosophical Problems of 'Cyber-
netics." This meeting was co-sponsored by
the Scientific Council on the Complex Prob-
lem "Philosophical Questions of Natural Sci-
ence," the Scientific Council on Cybernetics,
and the Party Committee, all of the Presidium
of the. Academy of Sciences, USSR. It was
attended by about 1,000 specialists, including
mathematicians, philosophers, physicists, en-
gineers, biologists, medical scientists, lin-
guists, psychologists, and economists, from
many cities of the USSR.3 Participants in-
cluded such prominent cyberneticians* as
A. N. Kolmogorov, V. M. Glushkov, A. I. Berg,
P. K. Anokhin, A. V. Napalkov, A. A. Feld-
baum, A. A. Lyapunov, S. V. Yablonskiy, I. B.
Novik, and Yu. Ya. Bazilevskiy. Of the 10
reports presented, six were closely related to
the problem of artificial-intelligence. A re-
port of this conference stated that "the prob-
lem most animatedly discussed was the most
disturbing of all, the problem of the techno-
logical operation of complicated psychic proc-
esses�the problem most frequently desig-
nated by the short and convenient although
not, in our view, entirely correct formula of
'can a machine think?' " 5
The discussion brought out three questions
as approaches to this problem. First, Is it
possible in general to reproduce with models
the complex intellectual activity of man?
Second, Can a machine surpass man in the
realm of intellectual activity and particularly
in the performance of creative tasks, such as
the formulation of new problems? Third, Is
it possible in principle, to achieve the exist-
ence of consciousness in a machine similar
to that exhibited by man?
Discussion at the June conference of the
first question may be described as informa-
tional and confirmatory rather than argu-
*The term "cybernetician" is applied by' the
Soviets to those �practicioners of traditional disci-
plines who exhibit the common characteristic of
dealing with their research ,problems in the frame-
work and methodology of cybernetics.
rnentative. There was general agreement
that, in view of the replications of intellectual
and sensory functions of man. already
achieved, an affirmative answer to questions
about the reproduction of human cognitive
processes cannot be doubted. The second
question was discussed more vigorously. Life
scientists, represented by P. K. Anokhin, a
leading neurophysiologist, argued that the
potentialities of a machine are limited to solv-
ing problems assigned by man using algo-
rithms ,of decision which, man puts into the
machine. This position was countered by
V. M. Glushkov, a leading Ukrainian cyber-
netician, and A. A. Feldbaum, Doctor of Tech-
nical Science in Electronics and member of
the automation faculty of the Lenin Power
Institute. They pointed out, that there are
machines today which, in the process of solv-
ing one complicated problem, can independ-
ently pose and solve a series of autonomous
problems of a particular. character. Glush-
kov maintained that any form of human
thought can, in an informational model, 'be
reproduced in artificially created cybernetic
systems. He agreed, however, that by ,virtue
of historical necessity�the .fact that it was
precisely man who created .machines for his
use and -not the reverse�the destiny of man
will always be the more important in the
processes of thought and cognition. Thus
Glushkov's assessment is that a machine can
be "smarter" than one man, or even a group,
but it cannot be "smarter" than human so-
ciety as a whole.
Questions about machine ,conaciousness
were argued most sharply at the conference.
One group argued the "black box" approach
to solution of these questions. This viewpoint
holds that man judges the presence of con-
sciousness in other people by .analogy, i.e., by
observing the behavior manifested by others,
in response to inputs (stimuli) and by com-
paring such manifestations with his own be-
havioral responses of which he is consciously
aware. Therefore, if a machine faithfully
produces the same outputs as a man in re-
sponse to the same inputs, it can be analo-
gously inferred to possess consciousness ac-
cording to this view. On the other side it was
FOR OFFICIAL U3E ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
ARTIFICIAL-INTELLIGENCE. �RESEARCH
IN THE USSR
PROBLEM,
To assess artificial-intelligence researdh: in the USSR
CONCLUSIONS
1. In the Soviet Union, 'substantial govern-
mental and Party support And encouragement
are being given to extensive studies on the
theory of artificial intelligence Soviet sci-
entists of high caliber are conducting arti-
ficial-intelligence 'studies � and are exposing
young students to. the state_ of .the art in
this interdisciplinary �field, laying a sound,
foundation for further advances.
2. The importance which the Soviet regime
attaches to artificial-intelligence research is
attested by the unusual freedom of open -dis-
cussion allowed scientists Working in the field,
as reflected in the published literature.
3. Soviet research in the theoretical and
hardware aspects of artificial-intelligence is
now about as advanced .as US work and can
be expected to continue, at a rate equal to or
.greater than that observed In the West. Sig-
nificant. theoretical achievements within the
next 5, years are highly probable. When
theories are converted into designs, Soviet, en-
gineers probably will be able to produce the,
:equipment.
SUMMARY
After a relatively' late start, Soviet research-,
On artificial-intelligence related to the ulti-
mate development of decision-making ma-
chines now is about, on ,a par with similar US,
Work. The apparently greater rate of Soviet
progress- compared with that of the West is
attributable to the Magnitude' of official rec-
ognition and support of, artificial intelligence
'research,in. the 'USSR' Soviet offiCials con-
sider the development of decision-making ma-
chines-to be essential to the successful man-
agement of their increasingly complex eco,-
nornia. and social,system, and are giving sub-
stantial support and Party, encouragement to
artificial-intelligence studies.
lit)K uFFICIAL U3E -ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL USE ONLY
Soviet ideologists and scientists are discuss-
ing, from a wide range of viewpoints, the
fundamental philosophical problems about
the nature of intelligence and of man which
are raised by attempts ,to model decision-
making or intelligent 'behavior. The stric-
tures of dialectical materialist dogma are not
Inhibiting research in this area. In view of
the high prestige of the Soviet scientists tak-
ing part in these discussions and the publica-
tion in the Soviet scientific literature of the
opinions expressed, artificial-intelligence stud-
ies represent rare and significant examples
of intellectual freedom in the USSR.
Current Soviet work in the major subfields
of artificial-intelligence research includes in-
vestigation of techniques for machine search
of problem solutions, pattern recognition, ma-
chine learning, planning and .inductiOn in
machine systems, and brain modeling. Soviet
progress in research on machine techniques
for search, pattern recognition, and learning
compares favorably with US advances. Sci-
entists and engineers engaged in research on
pattern-recognition in the Soviet Union have
made original contributions to the theoreti-
cal basis for artificial pattern-recognition sys-
tems and have fabricated various types of
pattern-recognition devices.
Soviet research on simulation of such as-
pects of the cognitive process as learning and
induction at present consists mainly of study
of self-teaching and self-organizing systems.
In these fields Soviet scientists and engineers
have created models based on neuro-physio-
logical conditioned-reflex approaches, on psy-
chological learning, theory, and on automatic
control system theory.
Soviet study of the exceedingly complex
problems involved in incorporating techniques
of planning and induction in artificial prob-
lem-solvers is at a very early stage of devel-
opment. As yet, little progress has been made
in these fields in the West.
Research on the-modeling of brain processes-
is receiving much emphasis in the USSR.
Since 1955, laboratory experimentation on
models that simulate neural processes has in-
creased, and outstanding neuro-physiologists
and psychologists have begun to work closely
with mathematicians, computer specialists,
and electronics engineers on cybernetic stud-
ies of brain-modeling problems.
Seminars are regularly offered to expose
students to the state of the art in the fields
that are especially significant to the future
of artificial-intelligence research in the USSR.
These seminars are conducted by the leading
researchers in artificial-intelligence ,and are
thereby laying a .sound foundation for future
advances in a new and multidisciplinary field.
DISCUSSION
INTRODUCTION
Since the mid-1950's, the Soviet govern-
ment has increasingly emphasized the impor-
tance of a cybernetics research program for
achieving national and international goals.*
A large part of the theoretical research aspect
of this program has been focused on the de-
velopment of decision-making machines.
Realization of such machines -would assist the
Soviets in solving two types of fundamental
_
�AEL:. 82H/46 �, The Meaning of Cybernetics in
USSR" bernetic 30 ,F7
Cy s in the USSR Mar 64
problems:' (i) decision making on the basis of
incomplete data (sometimes referred to as de-
cision making under conditions of uncer-
tainty), and (ii) decision making in the pres-
ence of complete but overwhelming, quantities
of data.. Problems of the latter type are- by
definition beyond. the. capabilities of -human
- decision .makers. � Problems of the first type,
while inherently characteristic of human
thinking. activity; are rapidly extendingbeyond�the bounds of possible human solution
in the context of eccinomic management and
the
state administration.
FOR OFFICIAL USE ONLY
�
(b)(3)
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Decision-making machines Are, in essence,
..models of one variety or another of the human
problem-solving process. . Models of the hu-
man thinking- or problem-solving process are
-artificial 'simulations .Of the information flOw,
or information processing, procedures of the
human problem saver. The model :may be
mathematical, in which case it may be a pro-
gram for a universal (digital) computing ma-
chine, or it may be A piece of hardware; such
as. an .electronic circuitry analog. (Many de-
cision procedures already elaborated by arti-
ficial-intelligence research are not efficiently
modeled by: the single-track, sequential meth-
od' of operation of 'conventional digital com-
puters. This 'Would not be surprising in view
of the apparent parallel operation of infor-
ination:processing.in- the human brain. Such
routines can be modeled for demon-
stration purposes on a. digital computer,. but
- analog models of some sort will loom increas-
ingly important in the fabrication' of artificial-
decision, makers.) In any case, the,essential
feature of such 'a .model .is that for a given
Input of information :the model shall _produce
at least the same output (of information). as
does the natural process being modeled. Re-
search on such models is categorized in, the
United States under the 'rubric "artificial:4h-
Soviet research on 'artificial:intelligence has
'Mushroomed since 1956 along with other sub-
problems of, cybernetics. After the establish-
Merit in 1959 Of a national program for cy-
:bernetics; the direction of Soviet research. on
Artificial-intelligence became: one of the re-
sponsibilities of the Scientific Problem Coun-
cil on Cybernetics of the Presidium, Academy
of Sciences, USSR :The Cybernetic Machines
'Section of the ,Council appears' to, be the -,unit
. Which assigns, monitors, and integrates. Most
of this research work on a national scale.
_Actual .research is conducted at the labora-
tories and institutes' listed th the appendix.
'The. number of facilities engaged in this re-
search can be expected to grow as the cyber-
netic approach .continues to permeate other
areas of Soviet science and technology appli-
cable to communication and control :problems
arising in,production-industries, the economy,
law, 'government administration, and military
activities.:
_Although: much of the application-oriented
cybernetics 'research is at present directed to-
'ward the, realizatithi of more, sophisticated
, conventional systems for automatic control
'Of Machines, plants, Space research Vehicles,
or - even economic units, a large part! of the
theoretical research is related, directly orin-
directly, to the development of "thinking ,ma-
-chines," or "thinking cybernatons," as Soviet
� scientists often refer to them. Realization of
such machines will have immediate relevance
to information abstracting and retrieval, ma-
chine translation, and general problem solv-
ing, and eventual application to later-genera-
tion automatic control= systems.-
. A. A. Lyapunov; a leading Soviet Mathema-
tician and editor of Problems . :or Cybernetics,
has described the relationship of artifionil-M-
telligerice to general cybernetics � and the na-
ture of future control systems He .suggests
that algOrithrnization (mathematical:model-
ing) of thinking processes and Of control p-rdC7
esses is one of the Most: important aspects of
.6rbernetic. theory., Other' members of the
Scientific Council on: Cybernetics, in explain-
ing the importance of artificial intelligence,
state that "the basic products- -of radioelee-
tronics; are various, devices for Automatic reg-
ulation, monitoring, control, and communica-
tions. The brain, :Whin fulfills these 'func-
tions � in the living organism, ivOrks.inueh inore
reliably and productively than any present-
day radioeiedtronic -machine. . . . Some first
steps nave already been taken in the direction
of constructing practical automata analo-
gous to the. brain."
PHILOSOPHICAL ASPECTS -
� Reexamination of basic philbsophiCal atti-
tudes toward the nature of the brain, of mind,
and of man 'himself is taking place wherever
artificial-intelligence research is being con-
ducted. The Soviet Union is no exception.
Philosophical-theoretical questions' bearing on
the -simulation of intelligence, or intelligent
behavior; are being thoroughly 'aired and in,.
3
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL Uf3E ONLY
vestigated ,in Soviet scientific circles, outside
as well as within the context of the funda-
mentals of dialectical materialism.2-8
A highly significant discussion took place
during June 1962 at a conference in 'Moscow
on "The Philosophical Problems of Cyber-
netics." This meeting was co-sponsored by
the Scientific Council on the Complex Prob-
lem "Philosophical Questions of Natural Sci-
ence," the Scientific Council on Cybernetics,
and the Party Committee, all of the Presidium
of the Academy of Sciences, USSR. It was
attended by about 1,000 specialists, including
mathematicians, philosophers, physicists, en-
gineers, biologists, medical scientists, lin-
guists, psychologists, and economists, from
many cities of the USSR .3 Participants M-
chided such prominent cyberneticians* as
A. N. Kolmogorov, V. M. Glushkov, A. I. Berk,
P. K. Anokhin, A. V. Napalkov, A. A. Feld-
baum, A. A. Lyapunov, S. V. Yablonskiy, I. B.
Novik, and Yu. Ya. Bazilevskiy. Of the 10
reports presented, six were closely related to�
the problem of artificial-intelligence.4 A re-
port of this conference Stated that the prob-
lem most animatedly discussed was the most
disturbing of all, the problem of the techno-
logical operation of complicated psychic proc-
esses�the problem most frequently desig-
nated by the short and convenient although
not, in our view, entirely correct formula of
'can a machine think?' P5
The diRrussion brought out three questions
as approaches to this problem. First, Is it
possible in general to reproduce with models
the complex intellectual activity of man?
Second, Can a machine Surpass man in the
realm of intellectual activity and particularly
in the performance of creative tasks, such as
the formulation of new problems? Third, Is
It possible in principle, to achieve the exist-
ence of consciousness in a machine � similar
to that exhibited by man?
Discussion at the June conference of the
first question may be described as informa-
tional and confirmatory rather than argu-
� The term "cyberneticia.n" Is applied by the
Soviets to those practicioners of traditional disci-
plines who exhibit the common characteristic of
dealing with their research problems in the frame-
work and methodology of cybernetics.
mentative. There was general agreement
that, in view of the replications of intellectual
and sensory functions of man already
achieved, an affirmative answer to questions
about the reproduction of human cognitive
processes cannot be doubted. The second
question was discussed more vigorously. Life
scientists, represented by P. K. Anokhin, -a
leading neurophysiologist, argued that' the
potentialities of a machine are limited to solv-
ing problems assigned by man using algo-
rithms of decision which man puts into the
machine. This position was countered by
V. M. Glushkov, a leading Ukrainian cyber-
netician, and A. A. Feldbaum, Doctor of Tech-
nical Science in Electronics and member of
the automation faculty of the Lenin Power
Institute. They pointed out that there are
machines today which, in the process of Solv-
ing one complicated problem, can independ-
ently pose and solve a series of autonomous
problems of a particular. character. Glush-
kov maintained that any form of human
thought can, in an informational model, be
reproduced in artificially created cybernetic
systems. He agreed, however, that by virtue
of historical necessity�the fact that it was
precisely man who created machines for his
use and not the reverse�the destiny of man
will always be the more important in the
processes of thought and cognition. Thus
-Glushkov's assessment is that, a machine can
be "smarter" than�one man, or even a group,
but it cannot be "smarter" than human so-
ciety as a whole.
Questions about machine consciousness
were argued most sharply at the conference.
One group argued the "black box' approach
to solution of these questions. This viewpoint
holds that man judges the presence of con-
sciousness in other people by analogy, i.e., by
observing the behavior manifested by others
in response to inputs (stimuli) and by 'com-
paring such manifestations with his own be-
havioral responses of which he is consciously
aware. Therefore, if a machine faithfully
produces the same outputs as a man in re-
sponse to the same inputs, it can be analo-
gously inferred to possess consciousness ac-
cording to this view. On the other side it was
ron OFFICIAL USE ONLY.
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
AL
maintained by some at the conference that
consciousness necessarily includes a subjective
:element with a specific Character Which is the
result of the labor and social -relations in
Which people engaged during the process of
social evolution.
A More empirical approach ,to conscious
ness---was advanced at the conference by those
scientists Interested inn-modeling the structure
Which: embodies human consciousness, that is,
the brain. Although a structural model of
the human brain is, far beyond present tech-
nological and scientific -capabilities, the: ,pds-
sibility)of its future realization is considered
to be worth discussion by the Soviets. A. N.
.Kolmogoroy, world renowned mathematician,
is ',optimistic about the chances of success in.
brain-modeling ventures. He called for the
treeing of definitions of life .and thought from
arbitrary premises and for the :redefinition of
theSe, concepts along purely functional lines
Kolmdgorov believes that if the functional
point of view toward life And thought is sub-
scribed ,to, ;the conclusion is inevitable that
replication of the, organization of a system
can be accomplished, by organizing different
:elements into a new system which would
:have the essential traits And structure of the
system -being modeled. From this Kolmogo-
roy :concluded that:
A sufficiently Complete model Of a living being
'can in all justice be, called a living being and
the model of a thinking being, ,a thinking being:
It ,is important distinctly to ,understand that
Within the limits of a materialistic ideology there
do ,not -exist any kind of well-grounded, principal,
arguments against a positive answer to (this)
nuestion. This positive answer is the contem-
porary form of the attitude concerning the nat-
� ural origin Of life and the material basis of
-consciousness.
The extreme empirical view advanced by
the Solinogoroi school seems to be, in the,
ascendancy, For example, several points: of
view on the question, Can a machine thin.k?"
have been �reviewed by a group of authors in:
the Works of the Ka?an Aviation Institute, a
somewhat surprising ,source for discussions of
philosophical aspects of artificial-intelligence.6
The group did not find Convincing any of the.
,arguments -against the possibility of Creating
a thinking machine. The subject of the -neg:-
ative arguments considered by them ran the.
gamut from the algorithmic insolvability of
some problems, and the irreducibility of the
thinking process to a physical Operation, to
the impossibility of modeling the subjective-
psychological world of man. The Kazan
group points out in rejecting such arguments,
that cybernetics provides the first basis for
(i) uncovering the elements involved in con-
sciousness and cognition and (ii) . "resolving
positively the question of the possibility of
creating :a thinking machine." The Kazan
group does not recognize the brain as the only
highly organized material system in which
consciousness can develop, and forecasts that
highly organized material systems of another
type, in which consciousness develops, are pos-
sible and realizable.
V. M.`Glushkov, one of the most politically
powerful among Soviet cyberneticists,* sides
with Kolmogorov and the Kazan group. Re-
cently, in discussing the possibilities of ma-
chine intelligence, he contrasted cybernetic
systems with earlier mathematico-logical and
other formal language approaches to model-
ing the thought processes. He found signifi-
cant advances in the cYbernetic approach.
Glushkov would describe any control or cogni-
tive system as a cybernetic system which can
be analyzed as an abstract [informational]
model. For this purpose, both the input and
Output information, that. is, all of the infor-
mation which a system exchanges with the
outside world, can be conceived as being en-
coded in words of a given standard alphabet.
All of the activity of the cybernetic system
may thus be reduced to the transformation of
words in a standard alphabet. The study of
a given cybernetic system can be reduced
thusly to the determination of rules accord-
ing to which the indicated transformation
occurs. Glushkov noted that among these
rules there may be some which permit certain
V. M. Glushkov is Vice President of the Academy
of Seiences, Ukrainian Soviet Socialist Republic;
Director of the Institute of Cybernetics, in Kiev;
Chairman of the Cybernetics Council of the Ukrain-
ian,Academy; and Chairman of the recently created
Interdepartmental Council for the Introduction of
Mathematical, Methods and Electronic Computers
in the National Economy, which is under the State
Committee for Coordination of Scientific ,Research
of the Council of Ministers, USSR'.
FOR OFFICIAL USE ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL UCE ONLY
chance transformations, as well as some that
permit the altering of other rules of informa-
tion conversion in the course of time under
the influence of an infinite surrounding me-
dium. Appreciating the possible infinitude of
the system of rules defining the regularities
of the informational activity of the brain,
Glushkov believes that the modeling of a suffi-
ciently large number of the essential rules in
the brain system eventually will result in a be-
havior pattern of the, model (on the informa-
tional level) which will correspond to brain
activity.8 Glushkov has asserted that, since
modern electronic digital computers possess
"algorithmic universality* . . . it is theoreti-
cally possible to model [on the informational
level] any thought processes with the aid of
such machines." 8 P 10
At least as significant as the content of
these epistomological disputations is their
wide distribution in popular and Party media.
Typical of "thinking machine" discussions is
an article in Znaniye-Sila (Knowledge is
Power, a popular-science type of magazine).
It describes the parallel mode of operation of
the brain (carrying out several calculations
or decision procedures simultaneously) versus
the series (single-track) nature of contem-
porary electronic computing apparatus and
points out the advantage of the former in effi-
ciency and universality. It reports that So-
Viet researchers are studying the operating
principles of automata which work in parallel
instead of in series.�
Soviet research in artificial-intelligence is
motivated by the anticipated necessity of
using machines in place of,people in situations
where speed, complexity, or other character-
istics of control processes exceed the capa-
bility of man. The Soviet policy regarding
the use of "thinking machines" was expressed
in a recent edition of Kommunist. The de-
velopment of technology, with the increase in
speed and accuracy requirements of separate
production operations and the growth of the
That is, computers can perform any Information
transformation on the basis of a program (algo-
rithm) built out of their available elementary in-
structions, if these include rules which define chance
transformations' and instructions by which certain
alterations are made in the system of rules.
entire technological process as a whole, was
said to have begun to exceed, in most cases,
man's power to control them. The Kommu-
fist article concludes that it is necessary to
replace even the psychic activity of man in
such:cases with automatic control machines.'�
PRINCIPAL RESEARCH PROBLEMS
The "tent-like" character of artificial-intel-
ligence research has resulted in a somewhat
'chaotic state in this field of science.11-'8
There are several schools, each represented
by spokesmen as critical of other schools as
they are competent in the techniques of their
own. The result is a' lack of standardized cri-
teria for use in assessing Soviet research in
the field of artificial-intelligence. Thus, an
expert in one popular US approach, upon dis-
covering a lack of comparable work in the
USSR, will give a negative evaluation of So-
viet research in artificial-intelligence. On the
other hand, the opponents of that particular
US expert will argue that the absence of such
research in the USSR signifies that the So-
viets have withdrawn from a blind alley of
investigation. Many of these conflicts are
semantic; almost all of the approaches to
artificial-intelligence share a common set of
problems. When Soviet research on these
problems is compared with US approaches to
the same problems, regardless of "schools,"
the USSR and the US are found to be approxi-
mately on a par.
There are differences in emphases, however,
between Soviet and US approaches to these
shared problems. US scientists tend to em-
phasize mathematical or machine models of
human cognitive processes. Many Soviet re-
searchers on the other hand consider that the
human brain has reached certain limits in re-
gard to its capabilities for memory and oper-
ational speed after a long process of slow
evolutionary development." The resultant
qualities of the human brain, therefore, are
not believed by the Soviets to be equal to all
the tasks modern men must face. According
to them, a machine that might be a perfect
analog of the brain, for. instance, will not do
any better than the brain when faced with
'6 �Z011�AFF-lehirL-154SE-01=--
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
OFFICIAL UGE ONLY
such tasks. Therefore, if the ever, more 'com-
plex problems are to be solved, machine
"brains" ,surpassing those of men must be
.built. M. V. KelclySh, President of the Acad-
emy of Sciences, USSR, in making this point.
asserts: "We must copy nature's processes in
technology creatively rather than literally,
with full knowledge of nature and of tech-
nology so that we may select.techniques'which
will give us better results than those achieved
In nature." 15: The natural processes to be
Copied "creatively" in the ,Soviet research pro-
gram are the same processes investigated by
US' scientists of most schools concerned with:
artificial-intelligence . These are seateh,pat-
tern recognition, learning, planning, and. ,in-
duction16-
Search and Pattern Recognition
The first approach to -machine solution- of a
problein is search. Given s' a problem, a con-
temporary data-processing :machine in the
USSR as elsewhere, can search rapidly by trial
and ,error through a large number of possible
solutions, for a valid solution to the given
problem. Nevertheless, in solving complex
non-trivial problems, the number of possible,
solutions is so large that this trial-and-error
methodology becomes ,excessively time-con-
suming ' in ,practical operation.
Soviet scientists recognize that a large re-
dyction in search time, although bringing
some real problems into the, realm Of practi-
cal machine solution, could be achieved by
the introduction of pattern-recognition tech-
niqUes. The machine that is designed with
pattern recognition aids-to-search could clas-
sify problems into categories amenable to cer-
tain types of solutions. The Current state of
Soviet and Western research suggests that
such techniques are nearing practicability for
'more-and more complex patterns.
TheoretiCal studies for pattern-recognition
devices began M the USSR as early as
1953.17'2'1 The philosophical, physical; -and
psychological bases of perception were eXten--
sively discussed in Voprosy Psikholoyii (Ques-
tions. of Psychology), in 1959.2' More .re-
cently, Soviet .researchers have accomplished
considerable work On the specifics of mini-
mum descriptions of images that are required
for recognition :by artificial systems.22-27
The works. of E. L. 131okh, mathematician
at the Institute for Information Transmission
Systems and E. M. Braverman of the Insti-
tute Of Autothation, and Telemethanics are
notable ,among recent approaches to practical
solutions of recognition problems. Braver-
man has originated a -"compactness -hypothe-
SiS" !" as a theoretical ,basis for solution of
such problems. Several Soviet researchers
are using this theory as a basis' for develop-
merit' of specific recognition techniques. E, L.
'Blokh ,is using' certain operations to compute
the distance and, angle between elements, rep-
resenting various Patterns presented, in an
n-dimensional configuration space. This ap-
pears to be a promising mathematical Model-
ing approach to, a large class of pattern-rec-
ognition. problerris.29
, Patternrrecognition modeling studies are
being- supported by research on perception
from the psychological and physiological
points of view. One 'investigation involved
the establishment and development of percep-
tual activity ,in 3- to 6-year-old children. Eye
movements of the ,subjects Were observed and
recorded photographically as the children 'ex-
aMined, (for learning) and later recognized
,pictures presented to them. The eye move-
ments recorded were then .compared with
measures of the recognition ability of the, chil-
,dren at: different ages.� Another study was
devoted to -identification of objects in the
visual''system. Time, requited for subjects to
recognize- simple objects, formed of small
numbers of elements, correlated well with
an information theory Model for such recog-
nition that Was ,based on earlier findings on
:information collating :and processing activity
in: .the-,eye 'systenia'
'The "compaetness hypothesis" is formulated as
follows: Patterns presented to the artificial system
are characterized by a number of criteria equal to
n. The va.lues:of the n criteria are used to estab-
lish points in an n-dimensional configuration space.
Each point represents an individual pattern. All
points representing patterns which are similar, for
example, all letters "A," all figures "5 ,".all pictures
-of cats,. will tend to lie- in "compact"regions of-the
space, with relatively easily discernible separations
between regions .28
FOR OFFICIAL- USE 0
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
-FVFM:frlitehtL�ifSE-49?%-11--
Soviet work in the area of realization of
physical models to perform pattern recogni-
tion has been reported. At a 1957 Scientific-
Technical Conference on Cybernetics, one
reading device for an information machine
was described.32 At an All-Union Conference
on Machine Translation, held at Moscow State
University in May 1958, reports on principles
of constructing electronic reading devices
were heard, and a device "enabling the blind
to read ordinary typographic text" was de-
scribed." 34 (The latter device has been pic-
tured in the Soviet press, but its operating
principles were not described in publication.)
Hardware modeling of pattern-recognition
schemes is being conducted by A. A. Kharke-
vich, the well-known radio engineer, ,as well
as others.33-37 6' Some of this research is di-
rected toward the specific application of auto-
mating theinput of information into comput-
ing machines."-"
A quasi-topological method of distinguish-
ing and identifying letters has been develoPed
and realized in hardware by a group of re-
searchers of the Institute for Systems for
Transmission of Information (Moscow).42 "
This makes use of scanning the contour of a
letter with a light spot and identifying its
topological characteristics (ends of lines, and
junctions of lines), which are recorded in a
binary code. Since this will not separate all
Cyrillic letters, some of which are topologi-
cally identical, further geometric analysis is
used to analyze topologically redundant
groups. Such a scheme is basically sound in
theory ,and relatively easily realized in hard-
ware, but system "noise" (disconnected lines
Or smudged letters) may be hard to. deal with,
and no figures have been published on reli-
ability of recognition accomplished.
In_June 1960, the Scientific Council on Cy-
bernetics sponsored a seminar on reading de-
vices. This seminar considered the principles
of constructing such machines and creating
corresponding systems for coding the infor-
mation involved." Five different machines
under development for automatic pattern
recognition were described by V. M. Glushkov
(letter recognition by line scan and minimal
description); V. A. Kovalevskiy (image scan-
ning by following the outlines of letters);
A. D. Krisilov (identification of constant fea-
tures of letters by means of standard tele-
vision techniques); V. M. Tsirlin (the quasi-
topological method) and A. G. Vitushkin (a
computer manipulated' system for analyzing
Cyrillic letters which separates characteristic
features by means of vertical line scanning).
In addition, E. M. Braverman and V. S. Fayn
presented papers on recognition systems em-
ploying learning (that is, performing identi-
fication on the basis of criteria not given be-
forehand).
Studies on the mathematical �modeling of
recognition processes on electronic digital
computers have been conducted by M. M:
Bongard, an outstanding young biophysi-
cist A report of his in the Cybernetics
Council collection Biologicheskiye A,spekty Ki-
bernetiki�Sbornik Rabot (Biological Aspects
of Cybernetics�Collected Works) describes
the methodology and prospects of recently
begun research aimed, first, at bridging, the
gap between physiological study of optical re-
ceptor activity and the modeling of recogni-
tion, using a universal digital computer.'"
However,. Bongard alludes to the disadvantage
of this model in contrast to the parallel in-
formation processing employed in human rec-
ognition activity. He foresees, therefore, the
development of a "logic of recognition . . .
a logic such that it could be used in an analog
computer. In essence; such a machine will
be a model of part' of the human brain."
Principles for constructing a "universal
reading" machine have been developed by
V. M. Glushkov." This scheme uses a cath-
ode ray tube-type receptor, a computer to
control the trace and to compute the descrip-
tion of the pattern presented, and a tech-
nique of comparison against descriptions' pre-
stored in the memory for identifying patterns
presented. The author admits that the
scheme described is unnecessarily cumbersome
for the recognition of such simple stylized
patterns as digits or letters, but points to its
usefulness for "reading" complex contours or
!8 -FOR�Crirrfc'USE ONLY
,
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR. OFFICIAL U3E ONLY
Semitone pictures. Such a "universal read-
ing automaton" has been built 'at, the Corn-
pitting Center ,of the Academy of Sciences,
Ukrainian SSR, and is used in conjunction
with a "Kiev" computer.
A somewhat different "machine that reads"
has been developed at the Ukrainian Academy
Computer , Center under the direction of Can-
didate of Technical Sciences, V. A. Kovalev-
skiy.4" " This system reportedly failed to
,recognize only 2 of 35,000. numbers produced
with a portable typewriter; including half
printed and otherwise distorted saMples. In
using this method - for recognition, according
to KoValevaltiy, the maximum of the correla-
tion coefficient for an unknown image and of
each of the standard images is sought, the
latter images 'being .subjected to all possible
transformations. When this is done, all nor-
mally typed letters, as well as .most of those
,artificially marred, are correctly recognized
and identified, with the statistical error of
incorrect recognition not exceeding'10-4.
For 'solving a more general problem .of rec-
ognizing nonstandard letters and nunibers, an.
algorithm is being developed which is based on
dichotomy, that is, sequential division of the
set of all' images into two classes. Koyalev-
skiy believes that such an algorithm will make
,it possible to work with an alphabet contain-
ing many characters and will assure rapid
recognition with a comparatively small mem-
,ory capacity.
Learning, Induction, and Planning
Further improvement in machine problem-
solving efficiency Could' be accomplished With
the addition of a learning capability. The
machine would then be able to apply readily
its already proven Methods to the solution
Of. problems- that are new but similar to 'prob-
-lemS previouslyi encountered in the Machine's
experience. Radical reductions of search
time could be realized through the application
Of planning methods:. the machine would:sur-
vey and analyze the solution space and plan
the best way for its detailed examination.
Furthermore; to manage broad classes-of very
complex problems, the machine, as the
human, must construct and internalize a
model of its environment, that is, it must
employ some scheme of induction.
'Research on planning and induction in ar-
tificial systems is at a rather early stage of
development in the USSR, as it is in the West.
Progress is occurring, however, in fields con-
tributing to the development of machine
learning, induction and planning. Such sup-
porting research includes studies on informa-
tion theory, coding theory, brain modeling,
statistical decision theory, automata theory,
and heuristic programming. Pertinent So-
viet literature often treats these subjects as
conjoined in such studies as pattern recogni-
tion employing learning, other learning sys-
tems, self-organizing systems, or brain models.
Learning appears to be an essential charac-
teristic of more efficient and truly universal
pattern-recognition systems, just as it is of
more efficient problem-solving appa.rata in
general. Soviet researchers in the field of
learning systems like Braverman, Glushkov,
and Mark Ayzerman,* compare favorably
with their Western counterparts. 'Further-
more, they are working on essentially the
same types of studies: perceptron-type sys-
tems, algorithms for teaching the recognition
Of shapes, and computer programs for recog-
nition of pattern configurations." "-�3
Investigations of learning systems for rec-
ognition are being conducted at a variety of
Soviet scientific establishments. At the In-
stitute of Automation and Telemechanics in
Moscow, a machine was programmed to dis-
cern numerical figures written in different
handwritings. According to Soviet reporters,
in only a few cases did the machine give er-
roneous responses, even when confronted with
previouslcr unseen figures. The Institute of
Surgery of the Academy of Medical Sciences
is testing the hypothesis that a "compact
area" is formed in the brain of an animal or
a,human by variants, of asimilar image. The
Institute of Biophysics of the Academy of Sci-
*M. A. Ayzerman is a Doctor of Technical,Sciences
in mathematics and electronics at the Institute of
Automation and Telemechanlcs, Moscow.
_Eirau�oppiehort�ust�marsr-
9
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL USE �Nil"
ences is attempting to program a machine
that will identify indices of an image and,
on the basis of the indices, to recognize the
image. In each of the experiments, errors
were made by the machine. However, the So-
viets believe that the important fact is that
the machine is capable of accumulating ex-
perience, with the result that its qualifications
are increasing and its errors are gradually
decrea.sing.55
M. A. Ayzfrman's latest experiments involve
the teaching of a machine to recognize pat-
terns without a need for pre-introduced cri-
teria. Starting from a basis in Braverman's
"compactness hypothesis," Ayzerman devel-
ops two algorithms by which a machine can
"learn" to distinguish between the compact
areas representing different images in a con-
figuration space, thus separating and recog-
nizing the images. The first algorithm has,
the machine construct, one by one, random
hyperplanes whose only criterion is that they
separate points which the machine is told
(that is, training process) represent different
objects. Ending up the training phase with
a series of intersecting hyperplanes, the ma-
chine' then examines these and "washes out"
sections of planes which do not perform the
separating function, thereby leaving a series
of broken hyperplanes which effectively sepa-
rate areas containing points representing dif-
ferent images. In using the second algo-
rithm, the machine constructs .positive poten-
tial surfaces (functions) decreasing away
from, respectively, each point or set of points
representing images of the same object.
Identification of a test image (point) is ac-
complished by (first algorithm) determining
on which side of the hyperplanes the point
lies, or (second algorithm) determining which
potential ,surface (function) has the highest
value at the test point. Tests were conducted
with five digits (0, 1, 2, 3, 5), each written
160 different ways. Using 40 samples of each
digit for training, and 120 for test, the ma-
chine achieved 83-89 percent correct recog-
nition with the first algorithm. With "paral-
leling" of seven variations of a digit in the
training process, 98 percent correct response
was achieved. The training sequence filled
10
1,500-3,000 binary digits in machine memory.
The second algorithm was. tested using 10
samples of each digit for training and 150
for testing and resulted in 100 percent cor-
rect recognizance. Additionally, the second
algorithm was tested on the 10 digits from 0
to 9, with 10 samples for training and 140 for
test on each digit, and achieved 85'percent
correct response.28 56 57 Ayzerman's second
algorithm is very similar to that employed in
a US device now becoming operational for the
identification Of sonar contacts.
Pattern-recognition techniques are em-
ployed at the Institute of Surgery imeni Vish-
nevskii, Moscow, to achieve rapid assessment
of the area and seriousness- of burns." The
algorithm, for recognition of objects with
many parameters, employs learning. The
system is "trained" on case histories. When
vital information, such as burn area and lo-
cation and patient's age, is fed in, the com-
puter identifies and stores symptoms and
other factors. It also identifies objective cri-
teria for forecasting the outcome of the ill-
ness. The system was tested on additional
case histories with known outcomes, and the
prognoses in most cases agreed with the ac-
tual course and outcome of the injury.
Many of the Soviet attempts to realize mod-
els of learning, induction, and other aspects
of the cognitive process are carried on under
the classification of self-teaching, (or learn-
ing) systems, or of self-organizing systems.
A significant portion of research in these areas
has apparently ,not been published. In a
number of cases the titles of papers discussed
at meetings and seminars have been pub-
lished, but the contentsof the papers are un-
available to the West. Thus, a self-teaching
machine based on a program model was dis-
cussed-at the First All-Union Meeting on Com-
puter Mathematics and Computing Technol-
ogy held in 1959, but details have not been
circulated to the West. Other self-teaching
machines were alluded to (but only in gen-
eral terms) at cybernetics seminars at Kiev
and Moscow State Universities.55-"T Since its
inception in 1955, the latter seminar, con-
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
�FOR-IMPFMAL�USE-49N-irl
ducted by the outstanding cyberneticist A. A.
Lyapunov, has held biweekly sessions through-
out the school year on a wide range of sub-.
jects related to cybernetics.
Application of principles from automatic
_control theory �to sell-organizing systems
study is exemplified by the work of the well-
known control engineer and mathematician,
A. G. Ivakhnenko. He has surveyed and cate-
gorized various types of "learning" and "self-
learning" (that is, new information generat-
ing) systems and has related US to Soviet
work on some of these types. Iva,khnenko is
studying^ the application of the theory of in-
variance and the principles of combined con-
trol systems to the development of certain
(self) learning systems. Control principles
apply to the memory part of the system, that
Is, to the control of the selection and accu-
mulation of information in the memory.
Ivakhnenko is specifically interested in a per-
ceptron-type device, a scheme first developed
in the United States.�8" Ivakhnenko's per-
ceptron device apparently employs some vari-
ations and innovations in comparison to simi-
lar US devices.7�
Modeling the processes of instruction with
automatic systems was discussed in a 1962
collection on automatic regulation and con-
trol. Starting from the learning theories of
Thorndike, Gestalt psychology ^and_I. P. Pav-
lov, the authors discussed various machine
learning systems. Among the� Western and
Soviet systems discussed were the perceptron
and the approaches of the US scientists
Newell, Simon and Shaw, Gelernter and
Rochester, and 0. Selfridge, the UK scientist,
Andrew, and the Special Design Bureau of
Moscow Power Engineering Institute. The
Soviets view training as a process of changing
algorithms, and, propose that "a systern which
finds by means of automatic .search an algo-
rithm of action which, is successful from any
determined point of view and which was not
put into the 'system by man before the train-
ing process, should be called a learning sys-
tem." 71
Closely related to systems embodying self-
learning are those capable of self-organiza-
tion.* Soviet scientists evinced interest in
the theory of self-organizing systems as early
as 1959. In that year, S. N. Braynes and A. V.
Napalkov wrote on the subject for Voprosy
Filosofii (Questions of Philosophy). In that
study, the investigators related the develop-
ment of such systems to their work on con-
ditioned-reflex modeling. They foresaw the
realization of "an algorithm of operation for
self-organizing .cybernetic systems, ensuring
the formation of new programs for operation
without the undertaking of 'exhaustive search'
of all possible variants." 73
Considerable attention was devoted to self-
organizing systems at an All-Union Meeting
on Computer Mathematics and Computing
Technology (1959) and at a symposium on
Principles of Design of Self-Learning Systems
held in Kiev during 1961. Comparison of the
published papers from the latter symposium
with those given at the first US Interdisci-
plinary Conference on Self-Organizing Sys-
tems in 1959 reveals very similar tripic cover-
age and a similar level of achievement �re-
flected at the two conclaves. As of 1961, the
Soviets were 2 to 3 years behind the United
States in this particular area of artificial-in-
telligence research.74-78
Brain Modeling
Historically, there has been a large amount
of Russian neurophysiological research since
the early 19th century, but its ma.thematiciza-
ton is a recent innovation. Brain research
now is very much concerned with the algo-
rithmization and modeling of the'information
transactions which take place in living or-
ganisms. These studies play an important
role in cybernetics/artificial-intelligence re-
*M C Yovits, Chairman of the First and Second
Conferences on Self-Organizing Systems, Chicago,
1960 and 1962, considers these areas of artificial-
intelligence research to be of great significance. To
Yovits it appears that "certain types of problems,
mostly those involving inherently non-numerical
types of information, can be solved efficiently only
with, the use of machines exhibiting a high degree
of learning or self-organizing capability. Examples
of problems of this type include automatic print
reading, speech recognition, pattern recognition, au-
tomatic language translation, information retrieval,
and control of large and complex systems. Efficient
solutions to problems of these types will probably
require some combination of a fixed stored program
computer and a self-organizing machine.",72
,FOR OFFICIAL WE ONLY
11
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR VI' k ildAL Ubh ONLY
search and are conjoined with attempts to
simulate artificially, the pattern recognition,
learning, planning and induction processes.
Before research could begin in this new
field, the whole area of physiology and cyber-
netics had to be broken out of the restraints
of Pavlovian doctrine. The beginning of this
break was apparent in a 1955 review of the
subject by the well-known Soviet physiologist
P. K. Anokhih in Questions of Philosophy
(Voprosy Filosofii) .7� Laboratory experimen-
tation in the modeling of brain processes
began shortly thereafter. Some of the earli-
est work was reported at a Scientific-Techni-
cal Conference on Cybernetics, held at the
Laboratory of Electromodeling of the Acad-
emy of Sciences, USSR, in May 1957. L. I.
Gutennaakher, Director of the Laboratory,
described work on the electrical modeling of
certain mental work processes using "infor-
mation machines with large internal stor-
age." 8� Research into the structural rnake-
up of the human brain was discussed at the
Seminar on Cybernetics at Moscow State Uni-
versity in 1960.8'
After an artificial-intelligence slant to tra-
ditional neuroanatomy and neurophysiology
became evident in efforts to model the brain,
a new type of interdisciplinary scientist
emerged. A. V. Napalkov of the Faculty of
Higher Nervous Activity at Moscow State Uni-
versity could well be described as the first
of this new breed of physiologist-cyberneti-
cist. In early 1959 he co-authored, with a
medical doctor and an engineer, a study which
surveys cybernetics,and physiology in general,
including the theory of automata. Further-
more, these scientists describe the results of
studies on brain activity in terms of a search
for algorithms representing systems capable
of independent development of new programs
for their, operation, and those able to form
new behavior patterns on the basis of proc-
essing information accumulated earlier.
They-also described an artificial device which,
in a primitive way, simulates these learning
processes, that is, a "learning automaton,"
and which was developed at the Moscow Power
Engineering Institute in cooperation with the
life scientists."
12
More intensive investigations into the infor-
mation processing procedures of the brain,
still in terms of the development of chains of
conditioned reflexes, were described by Napal-
kov in 1960.83 In 1962, the researchers in the
Department of Higher Nervous Activity re-
ported findings which Showed increasing so-
phistication in the greater complexity of the
algorithms of information processing that
had been derived. By 1962, a much more, so-
phisticated "learning machine," based on the
algorithms defined by the neurophysiologists,
and exhibiting some capability at "self-organ-
ization" (that is, self-improvement) , had been
fabricated by the engineers at the Power En-
gineering Institute. This group worked with
the neurophysiological laboratory of S. N.
Braynes, co-worker and co-author with Napal-
kov.84
Soviet Bloc researchers are also investigat-
ing the mode of operation of brain processes
from the point of view of psychology. The
work of a Czech, E. Goias, on the conditions
of generalization in pattern recognition and
learning falLs into this class of research. His
experiments involved a statistical analysis of
the process of generalization as manifested
by subjects perceiving common elements
among sets of stimuli (objects) presented.
This study, clearly of a preliminary nature,
served only to demonstrate that wide varia-
tions characterize the conditions for general-
ization."
New centers for brain research along cyber-
netic lines are now being established at the
Brain Institute, Institute of Physiology, Insti-
tute for Information Transmission Problems
and at installations outside the Moscow-Len-
ingrad complex. At Kiev, for example, stu-
dents are offered the opportunity to obtain
training in the most advanced areas of arti-
ficial-intelligence research, and specifically
'brain modeling. In 1962, two seminars were
held under the auspices of the Cybernetics
Council of the Ukrainian Academy of Sci-
ences. The first, on "Automation of Thinking
Processes," was conducted by V. M. Glushkov,
chairman of the Council. This seminar cov-
ered: (i) the foundations and particularities
of thought processes that are characteristic of
-poi eeA ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL UBE ONLY
man in the creative sphere of his activity, and
the possibilities of their algorithmic descrip-
tion; (ii) modeling on, contemporary com-
puters of such processes as pattern recogni-
tion, recognition of concepts, identification of
meaningful sentences, deduction of logical
consequences, proving theorems, selections of
strategies in games, and composition of
music; (iii) learning as a basis for modeling
the mental activity of man; (iv) theory of
self-teaching systems and practical develop-
ment of algorithms incorporating learning;
and (v) correlations between precise (to the
degree possible) modeling of creative proc-
esses and the specifics of machine algorithms
simulating these processes.
The second seminar, at the Ukrainian Acad-
emy, was led by Doctor of Medical Sciences,
N. M. Amosov. Problems associated with bio-
cybernetics and the application of electron-
ics in biology and medicine were considered
at this seminar. Specific topics included (i)
application of information theory in biology
and medicine; (ii) principles of automatic
control_in biological systems and their peculi-
arities; (iii) some principles of coding infor-
mation in the nervous system; (iv) perception
and transformation in receptors and the cen-
tral nervous system; (v) contemporary hy-
potheses on the nature of nerve excitation
from the position of biocybernetics; (vi) some
questions of modeling elements of the central
nervous system; (vii) thinking and the psy-
chic activity of man; (viii) principles of form-
ing self-organizing neuron nets and bionics;
(ix) control of the processes of excitation and
inhibition in the central nervous system by
means of electrical and electromagnetic in-
fluences; and (x) cellular biology in the light
of biocybemetics."
FOR OFFICIAL USE ONLY
13
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
APPENDIX
Institutes and Scientists Associated with Research of
Artificial Intelligence Significance in the Soviet Bloc
Academy of Medical Sciences, USSR
Institute of 'the Brain:.
Glezer, V. D. Retinal activity in identification
Nevska:ya, A. V. (probably). Retinal activity in, identification
Seredinskii, A. V. (probably). Retinal activity in, identification
Tsukkerman, I. I. (Probably). Retinal' activity In identification
Institute 'of Surgery. imeni Vishnevskii: learning systems , for recogni-
tion (location of images in the brain) ; :pattern-recognition techniques -
for rapid assessment of burns
Brayries S. N Head of Neurophysiological Laboratory. Algorithms-
of conditioned reflex development': neurocybernetics; self,-organiz-
ing systems
Academy of Pedagogical Sciences, RSFSR
Institute of Psychology:
Leont'ev, A. N. Information processes in man
Moscow State Pedagogical Institute:
Grishchenko, N. M. Recognition of meaningful sentences
Scientific Research Institute of Defectology:
Muratov, R. S. Reading devices
Academy, of Scienees, USSR
'Computer Center:
Kozhukin, G. 'I. Self =teaching maehines
Institute of Automation and_Telernechanlcs: learning systems for..rec-
ognitIon� .,(machine to discern handwritten. numerical figures)
.Ay2,erman,.m. A. Learning-systeins for recognition
Bashkirov,. 0. A. Learning systems for recognition
BraVerman,_E. M. Learning systems for recognition; "compactness-
hypothesis"
Feldbaurn, A.-I. Machine intelligence
Muchnik, I. H. Learning .systems for recognition
Shtil'man, Ye. V. Modeling instruction process using psychological
learning theory.
-----P01�APIAtettt�USEZMY� 15
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
�Feft-eiftilettt�trft ONLY
Institute of Biophysics: mathematical modeling of recognition proc-
esses; learning systems for recognition (identification of images by
means of indices)
Bongard, M. M. Mathematical modeling of recognition processes;
learning systems for recognition
Maksimov, V. V. Learning systems for recognition
Petrov, A. P. Learning systems for recognition
Smirnov, M. S. Learning systems for _recognition
Vayntsvayg, M. N. Learning 'systems for recognition
Zenkin, G. M.. Learning systems for recognition
Institute of Philosophy:
Novik, I. B. Modeling information processes
Institute of Physiology imeni Pavlov:
Anokhin, P. K. At Cybernetics Laboratory. Physiology and
cybernetics
Laboratory of Electromodeling: site�of Scientific Technical Conference
on Cybernetics (1957)
Avrukh, M. L., editor of a VINITI publication. Reading devices
Gutennaakher, L. I. Director of Laboratory of Electromodeling.
Electrical modeling of thought processes; automating informa-
tion input
Kholsheva, A. F. Reading device
Stretsiura, G: G. Reading devices
Mathematics Institute and Computer Center (Novosibirsk) : self-teach-
ing machines
Kozhukin, G. I. Self-teaching machines
Mathematics Institute imeni Steklov:
Kolmogorov, A. N. Parallel-operating automata; modeling think-
ing beings
Lyapunov, A. A., editor, Problemy Kibernetilci. General cybernetics
Lyubimskii, E. Z. Reading devices
Of man, Yurii. Parallel-operating automata
Mathematics Institute imeni� Steklov, Leningrad Department;
ve.rshayskii, V. L Minimum'clescription of images required for arti-
ficial recognition; pattern recognition with learning
Party Committee of the Presidium: Co-sponsored conference on ''Phil-
osophical Problems of Cybernetics"
Scientific Council on Cybernetics: general coordination of cybernetics
research work; *sponsored seminar on "Reading Devices"; co-sponsored
conference on "Philosophical Problems of Cybernetics"
Parin, V, Chairman of Section on "Cybernetics and Living Nature"
(Bionics)
Prokhorov, A.
Scientific Council on "Philosophical Problems of Natural Sciences": Co-
sponsored conference on "Philosophical Problems of Cybernetics"
Academy of Sciences, Latvian 'SSR
Institute of Electronics and Computer Technology:
Dambitis, Ya. Ya. (probably). Self-organizing systems
Institute of Physics:
Shneps, M. A. Self-organizing systems
FOB OFFIrIAL TJSE ODTT Y
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
�E014-43F-Fgehirt-TISE-01511-7�
Academy of Sciences, Ukraine SSR,
Computer Center:
Glushkov, V. M. Machine intelligence; self-teaching/self-organizing
systems; pattern recognition with learning; reading devices; rec-
ognition of meaningful sentences
Kovalevskii, V. A. Pattern recognition with learning; reading de-
vices; automation of information input
Stognii, A. A. Recognition of meaningful sentences
Cybernetics Council: sponsored seminars on "Automation of Thinking
Processes" and biocybernetics
Glushkov, V. M., Chairman of Cybernetics Council. Machine intelli-
gence; self-teaching/self-organizing systems; pattern recognition
With. learning; reading devices; recognition of meaningful sen-
tences
Stognii, A. A., Scientific Secretary of Cybernetics Council. Recogni-
tion of meaningful sentences
Institute of Cybernetics:
Gluskov, V. M. Director. Machine intelligence; -self-teaching/self--
organizing systems; pattern recognition with learning; reading
devices; recognition of meaningful sentences
Institute, of Electrical Engineering:
Ivakhnenko, A. G. Control theory in, artificial intelligence; percep-
tion-type device
Mathematics Institute:
Amosov, N. M., Leader .of Seminar on Biocybernetics
,Kukhtenko, A. I. Self-organizing (control) systems
Other Institutes and Personnel, Associated with Artificial-Intelligence
First Moscow Medical Institute, Department of Physiology:
Anokhin, P. K., Head of Department of Physiology. Physiology and
cybernetics
Kazan Aviation Institute:
Borshche, V. B., published in Trudy Kazan Aviatsionnyi Institut
(Works of the Kazan Aviation Institute). Machine intelligence
Ii in V. V., published in Trudy Kazan Aviatsionnye Institut. Ma-
chine intelligence
Rokhlin, F. Z., published in Trudy Kazan Aviatsionnyi Institut. Ma-
chine intelligence
Kiev 'Computer Center:
Kondratov, A., associated with work at Kiev Computer Center; writer
on artificial intelligence
Kiev Institute of Civil Air Fleet imeni Voroshilov:
Kukhtenko, A. I. Self-organizing (control) systems
Laboratory (now Institute) for Systems for Transmission of Information:
Blokh, E. L. Minimum description of images required for artificial
recognition; "compactness hypothesis"
Garmash, V..A. Quasi-topological approach to recognition; reading
devices
8FFAL 17
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
ron OFFICIAL UCE ONLY
Kharkevich, A. A. Minimum description of images required for
artificial recognition
Kirillov, N. Ye. Automatic discrimination of speech sounds
Pereverzev-Orlov, V. S. (probably). Quasi-topological approach to
recognition; reading devices
Tsirlin, V. M. Quasi-topological approach to recognition; reading
devices
Latvian State University, Computer Center:
ArM', E. .L Self-teaching machines
Leningrad State University, Experimental Laboratory of Machine Trans-
lation:
Andreev, N. D. Reading devices
Military Air Engineering Academy imeni Zhukovskiy:
Chinayev, P.. I. Self-teaching/Self-organizing systems
Moscow Power Engineering Institute: brain modeling
Kushelev, Yu. N. Engineer. Neurocybernetics
Krug, G. K. Self -teaching. machines
Letskii, E. L. Self-teaching machines; neurocybernetics
Svechinskii, V. B, student. Neurocybernetics, modeling thought
processes
Moscow State University: site of continuing Cybernetics -Seminar
Leont'ev, A. N. Information processing in man
Lyapunov, A. A., coordinator of Cybernetics:Seminars. General cy-
bernetics
Moscow State University, Department of Higher Nervous Activity:
Chichvarina, N. A. Algorithms of conditioned reflex development;
neurocybernetics
Napalkov, A. V. Algorithms of conditioned reflex development;
neurocybernetics; self-organizing systems
Seinenova, T. P. Algorithm a of conditioned reflex development;
neurocybernetics
Shtil'man, Ye. V. Modeling instruction processes using-psychologi-
cal learning theory
Sokolov, Ye. N. Modeling perception
Turov, A. F. Algorithms of conditioned reflex development; neuro-
cybernetics
Voloshinova, Ye. V. Algorithms of conditioned reflex development;
neurocybernetics; modeling instruction processes using psycho-
logical learning theory
Order of Lenin Institute of Power:
Ferdbatun, A. I., Faculty of Automation-and Computer Technology.
Machine. intelligence
University imeni Palacki, Olomuc, Czechoslovakia, Chair of Psychology:
Goias, E. Generalization in pattern recognition and learning
Osladilova, D. Generalization -in pattern recognition and learning
Valousek, C. Generalization in pattern recognition and learning
Zaporozhe Oblast Psychiatric Hospital:
Gasul', Ya. R. Modeling thought processes
18 FOR OFFICIAL USE ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFTCT AT. USE ONT y
Additional Personnel, Associated with'Artificial-Intelligence
Biryukov, B. V.,,contributor to Problem Kirbernetiki
Blinkov, S. M. Structure of the brain
Fain, V. S. Minimum description of images required for artificial
recognition; learning systems for recognition; automating infor-
mation input
Faticin, L. V., contributor to Voprosy Psikhologii (Questions of Psy-
chology)'. Philosophical problems of cybernetics; automatic dis-
crimination of speech sounds
Gutchin, Izmir, contributor to Zarya Vostoka (Dawn of the East) ,
Tbilisi, Georgian SSR
Kamynin, S. S. Reading devices
ICrisilov, A. D. Reading devices
Kubilyu.s, I., Prof., contributor to Kommunist (Vil'nyus). General
cybernetics
Mayzel', N. I., contributor to Voprosy Psikhologgi (Questions of
Psychology) ; philosophical problems of cybernetics
Mitulinskii, Yu. T. Minimum description of images required for
artificial recognition
Petrenko, A. I., contributor to lzvestiya Vuz (News of Higher Edu-
cational Institutions). -Reading devices
Pollakov, V. G., contributor to lzvestiya AN SSSR (News of the
Academy of Sciences, USSR),. Reading devices
Rozhanskaya, E. V. ,Mathematical theory of intelligibility (recogni-
tion)
Rybak, V. I. (probably Computer Center AS, T.JkSSR). Pattern rec-
ognition with learning
Saenko, G. I., editor of VINITI publication. Reading devices
Saparina, Ye. Brain modeling
Semenova, T. N. Pattern recognition .with leanling
Sindilevich, L.,Mweditor of VINITI publication. Reading devices
Sokolovskii, V. ,A. Minimum description of images required for
artificial recognition
Svechnikov, S. V., contributor to lzvestiya VUZ (News of Higher Ed-
ucational Institutions) . Reading devices
Tarakanov, V. V. Perceptual activity in man
Tiukhtin, V. S. Theory of Images (and Perception)
Tsemel', G. L Automatic discrimination of speech sounds
Vasil'ev, A. M., editor of VLNITI publication. Reading devices
Vitushkin, A.- G. Reading devices
Wang, Chih-ch'ing. Perceptual activity in man
Yeliseyev, V. K. Modeling ,recognition process
Zinchenko, V. P. Perceptual activity in, man -
19
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL USE OM,
REFERENCES
1. Lyapunov, A. A. "Certain General Prob-
lems of Cybernetics," Problemy Kiberne-
tiki (Problems of Cybernetics), v 1, Mos-
cow, 1958, p 5 (Pergamon Press, London,
1960)
2. Parin, V. and A. Prokhorov. "Uncovering
the Secrets of Living Nature," Moscow,
Pravda (Truth), 24 Mar 63
3. Biryukov, B. V. "Conference on Philo-
sophical Problems of Cybernetics," Pro-
blemy Kibernetiki (Problems of Cyber-
netics), v 9, Moscow, 1963, p 349 (JPRS:
21,448, 14 Oct 63, p 611)
4. Papers related to artificial-intelligence
presented at the Conference on Philo-
sophical Problems orCybernetics as listed
in 3. above:
Leont'ev, A. N. "On Some Particulars of
the Processing of Information by Man"
Kolmogorov, A. N. "Life and Thinking.
from the Point of View Of Cybernetics"
Lyapunov, A. A. "On Control Systems in
Living Nature and General Understand-
ing of Living Processes"
�Glushkov, V. M. "Thinking and Cyber-
netics"
Novik, I. B. "On the Nature Of Infor-
mation and the Peculiarities of Cyber-
netic Modeling"
Fel'dbaum, A. I. "Instruction Processes
for People and for Automata"
5. Mayzer, N. I. and L. V. Fatkin. "Confer-
ence on Philosophical Problems of Cyber-
netics," Voprosy Psikhologii (Questions of
Psychology), Moscow, no 5, 1962 (JPRS:
� 17,574, 12'Feb 63)
6. Win, V. V., Borshche, V. B., and Rokhlin,
F. Z. "Can a Machine Think?", Trudy.
Kazansk Aviatsionnogo lnstituta (Works
of the Kazan Aviation Institute), Issue
65, 1961 (Referativnyi Zhurnal�Mate-
matika) (Reference Journal--Mathema-
tics), no 2, Feb 63, Abst 2 V213)
7. ,Gasul', Ya. R. "On the Possibility of Ob-
taining a Model for Thought Functions,"
Kiev, Fiziologicheskiy Zhurnat (Physio-
logical Journal), v 8, no 6, Nov-Dec 62
8. Glushkov, V. M. "Thought and Cyber-
netics," Voprosy Filosofii (Questions of
Philosophy), Moscow, v 17, no 1, 1963, p
36-48 (JPRS: 18302, 22 Mar 63)
9. Kondratov, A. "Automata, Thinking and
Life," Znaiziye-Sila (Knowledge is Pow-
er), Moscow, no 6, 1962. (JPRS: -15,851, 23
Oct 62)
10. Kubilyus, I. "Cybernetics and Life,"
Virnyus, Kommunist (Communist), no 4,
Apr 63
11. Minsky, Marvin. "Steps Toward Artificial-
Intelligence," Proceedings of the IRE, v
49, no 1, Jan 61
12. . "A Selected Descriptor-Indexed
Bibliography to the Literature on Artifi-
,cial-Intelligence," IRE Transactions on
Human Factors in Electronics, 1961
13. Gruenberger, Fred. "Benchmarks in Ar-
tificial-Intelligence," Datamation, v 8, no
10, Oct 62, p33-35
14. Berg, A. "Cybernetics and Its Domain,"
Vechernyaya Moskva, 11 Nov 63, p 2
15. Keldysh, M. V. "The Development of Cy-
� bernetics is the Concern of Scientists
from Various Specialties," Technical Cy-
bernetics, no 3, 1963 (JPRS: 21,100, 16
Sep 63, p 4)
16. Lincoln Laboratory Library. 8th refer-
ence bibliography, Artificial-Intelli-
gence--Soviet Bloc, Morris D. Friedman,
compiler, 27 Dec 61
17. 'Rozhanskaya, E. V. "On the Question of
the Mathematical Foundation of the
Theory of Intelligibility," Trudy, Komitet
po Akustiki (Works, Committee on Acous-
tics), no 7, 1953
FOR OFFICIAL VOE ONLY
21
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
18. Kamynin, S. S. and Lyubimskiy, E. Z.
"Principles of Reading Symbols Using
Automatic Machines," at Scientific-Tech-
nical Conference on Cybernetics (1957)
as reported in Pro blemy Kibernetiki
(Problems of Cybernetics), v 1, 1958, p
266 (Pergamon Press, London, 1960)
19. Itharkevich, A. A. "Pattern Recogni-
tion," Radiotekhnika (Radio Engineer-
ing), v 14, no 5, 1959
20. Sokolov, E. N. "A Probability Model of
Perception," Voprosy ,Psikhologii (Ques-
tions of Psychology), v 6, no 2, 1960
21. Tiulaitin, V. S. "K Probleme Obraza"
(On the Problem of Image), Voprosy Filo-
sofii (Questions of Philosophy), no 6, 1959
22. Blokh, E. L. "Toward the Question of
Minimum Description," Radiotekhnilca
(Radio Engineering), v 15, no 2, Feb 60
23. Fayn, V. S. "On the Quantity of Coordi-
nate Descriptions of Images in Sys-
tems for Visible Pattern Recognition," lz-
vestiya AN SSSR, OTN , Energetika i Avto-
matika (News of the Academy of Sci-
ences, USSR, Department of Technical
Sciences, Power Engineering and Auto-
mation), no 2, 1960
24. . "On an Abbreviated Inscription
of Absolute Descriptions of Images," Iz-
vestiya AN SSSR, OTN , Energetika i Avto- 31.
nurtika, no 1, 1961
25. Varshayskii, V. L and Sokolovskii, V. A.
"On the Problem of Recognizing Configu-
rations," Kibernetilca i Elektronno-Vy-
chisliternaya Tekhnika po Mat erialam'
XVI Nauchno-Tekhnicheslcoi Kcmferen-
tsii (of Nauchno-Tekhniche.skoye Ob-
shch estvo Radiotekhnzki i Elelctrosvyazi
tin A. S. Papaya) (Cybernetics and Elec-
tronic-Computing Technology�Materials
of the 16th Scientific-Technical Confer- 32.
ence�of the Popov Society), Moscow/
Leningrad, Gosenergoizdat, 1962, p 12
26. Mitulinskii, Yu. T. "Recognition of Digi-
tal Symbols," Voprosy Vychisliternoy
Tekhniki (Questions of Computer Engi-
neering), Kiev, Gostekhizdat UkSSR, 1961
(JPRS: 16,490, 3 Dec 62, p 21)
22; r OR �MGT
27. Kharkevich, A. A. "Selection of Criteria
in Mechanical Recognition," lzvestiya
AN SSSR, OTN, Tekhnicheskaya Kiber-
netiki (News of the Academy of Sciences,
USSR, Department of Technical .Sciences,
Technical Cybernetics), no 2, Mar/Apr
63 (JPRS: 19,966, 1 Jul 63, p 1)
28. Ayzerman, M. A. "Experiments on
Teaching Machines to Recognize Visual
Images," Biological Aspects of Cyber-
netics�Collection of Works, Department
of Biological Sciences, Scientific Council
on the Complex Problem "Cybernetics,"
Academy of Sciences USSR Publishing
House, Moscow, 1962 (JPRS: 19,637, 11
Jun 63, p 242)
29. Blokh, E. L. "Some Approaches to a
Practical Solution of the Problem of Rec-
ognition," lzvestiya AN SSSR, �OTN,
Tekhnicheslcaya Kibernetiki (News of the
Academy of Sciences, USSR, Department
of Technical Sciences, Technical Cyber-
netics), no 2, Mar/Apr 63 (JPRS: 19,966,
1 Jul 63, p 14)
30. Zinchenko, V. P., Wang Chih-ch'ing and
V. V. Tarakanov. "The Establishment
and Development of Perceptual Activity,"
Voprosy Psikhologii (Questions of Psy-
chology), Moscow, no 3, 1962 (JPRS:
16,404, 29 Nov 62, p 1)
Glezer, V. D., A. A. Nevskaya, A. V.
Seredinskii and I. I. Tsukkerrnan. "On
the Identification of Objects in the Visual
System," Biologicheskiye Aspekty Kiber-
netiki--Sbornik Rabot (Biological Aspects
of Cybernetics�Collection of Works), De-
partment of Biological Sciences, Scien-
tific Council on the Complex Problem
"Cybernetics," Academy of Sciences USSR
Publishing House, Moscow, 1962 (JPRS:
19,637, 11 Jun 63, p 225)
Avrukh, M. L., A. F. Kholsheva., and G. G.
Stretsiura. "Reading Device for an In-
formation Machine," Scientific-Technical
Conference on Cybernetics (1957) (see 18
above)
33. Andreev, N. D. 'Principles of Construc-
tion of Electronic Reading Devices," at
All-Union Machine Translation Confer-
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
-FOR-CforflefIrL�tfeE-eNTAr�
ence (1958), as reported in Problemy,
Kibernetiki (Problems of Cybernetics),
v2, 1959, p307
34. Muratov, R. S. "Device for Reading of
Ordinary Typographic Text by the
Blind," All-Union -Machine Translation
Conference (1958) (see 33 above, p 308)
35. Kharkevich, A. A. "On Principles of
Constructing Reading Machines," Radio-
tekhnika (Radio Engineering), v 15, 'no
2, Feb 60
36. Fyn, V. S. "On Principles of Construct-
ing Machines for Recognition of Pat-
terns," Radiotekhnika (Radio Engineer-
ing), �v 15, no 3, Mar 60
37. Petrenko, A. I. and Svechnikov, S. V.
"Main Trends in the Development .of
Reading Automata," 'Izvestiya Vysshikh
Uchebnykh Zavecleniy, MVySSO, Radio-
tekhnika (News of Higher Educational
Institutes, Ministry of Higher and Spe-
cialized Secondary Education, Radio En-
gineering), v 4, no 3, 1961 (JPRS: 11,524)
38. Fayn, V. S. "On Automation of the In-
put of Certain Kinds of Data into Com-
puters," Avtomatika i Telemekhanika
(Automation and Telernecha.nics), v 22,
no 4, 1961
39. Gutenmakher, L. I. "Elektronnye Infor-
rnatsionno-Logicheskie Mashiny" (Elec-
tronic Information-Logical Machines),
2nd Ed, Moscow Academy of Sciences
Press, 1962, p 36-49
40. Kovalevskii, V. A. "Construction of Read-
ing Automata for Electric Digital Ma-
chines," at Kiev Cybernetics Section, as
reported in Problemy Kibernetiki (Prob-
lems of Cybernetics), v 6, 1961
41. Biological Aspects of Cybernetics�Collec-
tion of Works (Biologicheskiy Aspekty
Kibernetiki�Sbornik Rabot), Department
of Biological Sciences, Scientific Council
on the Complex Problem "Cybernetics,"
Academy of Sciences USSR Publishing
House, Moscow, 1962 (JPRS: 19;637, p
331-334)
42. Pereverzev-Orlov, V. S. and Poliakov, V. G.
"On the Construction of Reading Ma-
chines," lzvestiya AN SSSR, OTN, Ener-
getilca i Avtomatika (News of the Acad-
emy of Sciences, USSR, Department of
Technical Sciences, Power Engineering
and Automation), no 3, 1961, p 110-112
43. . "Universal Automat for Reading
Printed Text," in Chitayushchiye Ustroi-
stva (Reading Devices), Moscow, 1962, p
73-82 (JPRS: 18,119, 13 Mar 63, p 169-
199)
44. Garmash, V. A. "Seminars on Reading
Devices," Vestnik AN SSSR (Herald of
the Acaderily. of Sciences, USSR), v 30,
no 9, 1960-
45. Bonga.rd, M. M. "Recognition Modeling
on a Computer," at Symposium on Prin-
ciples of Design of Self-learning Systems
(1961) .as reported in Problemy Kiber-
netiki (Problems of Cybernetics) , v 7,
Moscow, 1962, p'231-32
46. "Modeling Recognition Proc-
esses," at Cybernetics Seminar, Moscow
'State University, 1961, as reported in
Problemy Kibernetiki (Problems of Cy-
bernetics), no 7, p 233
47. . "Modeling the Process of In-
structing for Recognition on a Universal
Computer," Biologicheskii Aspekty Ki-
bernetiki--Sbornik Rabot (Biological As-
pects of Cybernetics�Collection of Works)
(see 41 above) -(JPRS: 19,637, p 261)
48. Glushkov, V. M. "Designing Principles
of a Universal Reading Automaton," Av-
tornatylca (Automation), no 1, Kiev, 1962,
p 55-64 (JPRS: 15,303, 17 Sep 62)
49. "A Machine That Reads," Pravda Ukrainy
(Ukraine Pravda), 12 Jan 62 (le1l.)-TT-
62-1205)
50. Yeliseyev, V. K. and Kovalevskiy, V. A.
"Study of an Algorithm for the Recogni-
tion of Machine-Written Symbols,"
Zhurnal Vychislitel'noy Mat ematiki i
Matematicheskoi, Fizilci (Journal of Com-
puter Mathematics and Mathematical
Psysics), v- 2, no 5, Sep/Oct 62 (JPRS:
16,343; 26 Nov 62)
51. Academy of Sciences, USSR, Institute of
Scientific Information, Avrukh, M. L.,
rem CIPPICIAL USE ONLY
23
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL UCE ONLY
Vasirev, A. M., Saenko, G. I., and Sinde-
levich, L. M. (eds.), Reading Devices, Mos-
cow, 1962, 188 pp (JPRS: 18,119)
52. Braverman, E. M. "Some Problems of
Design of Machines Which Classify Ob-
jects According to Criteria Previously
Unknown," Avtomatika i Telemekhanika
(Automation and Telemechanics), Mos-
cow, no 10, 1960, p 1375-1386 (JPRS:
9028, 27 Apr 61)
53. Glushkov, V. M., Kovalevskiy, V. A., and
Rybak, V. I. "On One Algorithm for
Teaching the Recognition of Patterns,"
at Symposium on Principles of Design of
Self-learning Systems (1961) (see 75 be-
low)
54. Semenova, T. N. and Varshayskii, V. I.
"A Learning Program for Recognition of
Configurations," at Symposium on Prin-
ciples of Design of Self-learning Systems
(1961) (see 75 below)
55. Gutchin, Izrail'. "Animals and Ma-
chines Cybernetics Performance," Zarya
Vostoka (Dawn of the East), Tbilisi, 28
Jul 62, p 4
56. Ayzerrnan, M. A. "Experiments with
Teaching a Machine to Recognize Pat-
terns Without the Introduction of Their
Properties [Descriptions] into the Pro-
gram of the Machines," Cybernetics Sem-
inar at Moscow State University, 1962, as
reported in Problemy Kibernetiki (Prob-
lems of Cybernetics), v 9, Moscow, 1963,
p345
57. "They Teach a Machine to Rec-
ognize Visual Images," Nauka i Zhizn
(Science and Life), no 12, Dec 62, p 34-39
(JPRS: 18,126, 14 Mar 63)
58. Varshayskii, V. I. and Semenova., T. N.
"Learning Program for Recognition of
Configurations," Kibern.etika i Elektrcm-
no-Vychislitel'naya Tekhnika (Cybernet-
ics and Electronic-Computing Technol-
ogy) (see 25 above)
59. . "Teaching the Identification of
Configurations Formed of the Simplest
Signs," at Symposium on Principles of
Design of Self-learning Systems (1961)
(see 75 below)
60. Glushkov, V. M., Kovalevskii, V. A., and
Rybak, V. I. "Algorithm of . a Learning
Machine for Recognition of the ,Simplest
Geometric Figures," at Symposium on
Principles of Design of Self-learning Sys-
tems (1961) (see 75 below)
61. Glushkov, V. M., Grishchenko, N. M., and
Stognii, A. A. "Algorithm for the Recog-
nition of Meaningful Sentences," at Sym-
posium on Principles of Design of Self-
learning Systems (1961) (see 75 below)
62. Tsemel', G. I. "Automatic Discrimina-
tion of Invariant Criteria of Occlusive
(Plosive) Sounds by Means of Clipped
Speech Signals," lzvestiya AN SSSR,
OTN, Energetika Avtomatika (News of
the Academy of Sciences, USSR, Depart-
ment of Technical Sciences, Power Engi-
neering and Automation), no 5, 1960
63. Kirillov, N. Ye. and F'atkin, L. V. "Ex-
periments in the Recognition of Speech
Sounds by Automatic Machines," Vo-
prosy Psikhologii (Questions of Psychol-
ogy), Moscow, no 3, May/Jun 62, p45-55
(JPRS: 15,745, 17 Oct 62)
64. Control, London, v 7, no 65, Nov 63, p 262
65. Kozhukin, G. I. "A Programmed Model
of a Self-teaching Machine," All-Union
Conference on Computing Mathematics
and Computer Technology (1960), as re-
ported in Problemy Kibernetiki (Prob-
lems of Cybernetics), v 5, Moscow, 1961
66. Ar, E. I. "On Self-Teaching Machines,"
Cybernetics Seminar at Moscow State
University, 1960, as reported in Problemy
Kibernetiki (Problems of Cybernetics),
Moscow, no 5, 1961
67. Glushkov, V. M. "Principles of Con-
structing Self-teaching Systems," Kiev
Section on Cybernetics, as reported in
Problemy Kibernetiki (Problems of Cy-
bernetics), Moscow, no 6, .1961,'.p 301
68. Rosenblatt, F. "Perceptual Generaliza-
tion Over Transformation Groups" in
Self-Organizing Systems, Yovits, M. C.
and Cameron, S. (eds.), N. Y., Pergamon
Press, 1960, p 63
24 FOR OFFICIAL UCE ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
FOR OFFICIAL USE 0
69. "A Comparison of Several Per-
ceptron Models" in Self-Organizing Sys-
tems 1962, Yovits, M. C., Jacobi, G. T.,
and Goldstein, G. D. (eds.), Washington,
Spartan Books, 1962, p 463
70. Ivakhnenko, A. G. "Application of the
Invariance Theory and Combined Control
in the Synthesis and Analysis of 'Learn-
ing' Systems," Avtomatyka (Automation),
Kiev, no 5, 1961, p 3-12 (FTD-TT-62-75,
30 Mar 62)
71. Voloshinova, Ye. V. and Shtil'man, Ye. V.
"On the Modeling of the Processes of In-
struction in Automatic Systems," in
Sbarnik: Avtomaticheskoye Regiairova-
niye i Upravleniye (Collection: Automatic
Regulation and Control) Moscow, Acad-
emy of Sciences, USSR, 1962 (from Refe-
rativnyi Zhurnal�Matematika (Refer-
ence Journal�Mathematics), no 2, Feb
63, Abst 2 V523)
72. Yovits, M. C. Preface to Proceedings of
the First Conference (1960), in Self-Or-
ganizing Systems 1962, op. cit., p viii
73. Braynes, S. N. and Napalkov, A. V.
"Nekotorye voprosy teorii samoorganizyu-
shchilchsya sistem" (Some Questions of
the Theory of Self-Organizing Systems),
Voprosy Filosofti (Questions of Philos-
ophy), Moscow, no 6, 1959, p 148-154
74. Darnbitis, Ya. Ya. and Shneps, M. A.
"On Self-Organizing Systems," All-Union
Conference on Computing Mathematics
and Computer Technology (1960) (see 65
above)
75. "Principles of Design of Self-learning
Systems," Kiev, Gostekhizdat, 1962
(JPRS: 18,181, 18 Mar 63) and reported
in Problemy Kibernetiki, v 7, Moscow,
1962, p 231-232
76. Chinayev, P. I. "Self-learning--One of
the Basic Means of Development of Self-
regulating Systems," at Symposium on
Principles of Design of Self-learning Sys-
tems (see 75 above)
77. Kukhtenko, A. I. "On Self-Organizing
Control Systems," at Symposium on
Principles of Design of Self-learning Sys-
tems (1961) (see 75 above)
78. "Self-Organizing Systems," Yovits, M. C.
and Cameron, S. (eds.), op. cit.
79. Anokhin, P. K. "Fiziologiya i Kiber-
netiki" (Physiology and Cybernetics),
Voprosy Filosofti (Questions of Philos-
ophy), no 6, 1955
80. Gutenmakher, L. I. "Elektricheskoe
modelirovanie nekotorykh protsessov um-
stvennogo truda s pomoshch'yu informa-
tsionnykh mashin s bol'shoi vnytrennei
pamyat'yu" (Electrical modeling of some
mental work processes using information
machines with large internal memory),
Scientific-Technical Conference on Cy-
bernetics (1957), as reported in Problemy
Kibernetiki (Problems of Cybernetics),
V 1, 1958, p 266 (Pergamon Press, London,
1960)
81. Blinkov, S. M. "On the Structure of the
Brain," Cybernetics Seminar at Moscow
State University, 1960, as reported in
Problemy Kibernetiki, v 5, Moscow, 1961
82. Braynes, S. N., Napalkov, A. V., and
Svechinskiy, V. B. "Scientific Notes
(Problems of Neuro-cybernetics)," Mos-
cow, 1959, 109 pp (JPRS: 5880, 18 Nov
60)
83. Napalkov, A. V. "Nekotorye Printsipy
Raboty Golovnogo Mozga" (Some Prin-
ciples of Operation of the Brain), Pro-
blemy Kibernetiki (Problems of Cyber-
netics), v 4, Moscow, 1960, p 183
84. . "Information Processing by the
Brain," in Biologicheskiye Aspekty Kiber-
netiki (Biological Aspects of Cybernetics),
Academy of Sciences USSR Publishing
House, Moscow, 1962 (JPRS: 19,637, 11
Jun 63, p 174)
85. Golas, E. "Study of the Conditions of
Generalization," Voprosy Psikhologii
(Problems of Psychology), Moscow, no
3, 1962 (JPRS: 16,404, 29 Nov 62, p 55)
86. Stognii, A. A. "On Coordination of Sci-
entific-Research Work on Cybernetics in
the Ukraine," Problemy Kibernetiki
(Problems of Cybernetics), v 9, Moscow,
1963, p 342 (JPRS: 21,448, 14 Oct 63, p
598-600)
25
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
_EOR-OMQAL--USEeNVI'
FOR OFFICIAL USE ONLY
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
Ii
�
osi - s AA -37
Sender Will Will Check Classification Top and Bottom
CONFIDENTIAL I SECRET TOP SECRET
MEMCRANDUM
TO : Chief, OSI Staff
ATTENTION : Editorial Section
FROM : Chief, Life Sciences Division/OSI
Date may 1964
SUBJECT WP 561107 9 Artific-ial-Intelligence Research in the USSR
ATTACHMENTS: 3 Copies of Subject Manuscript
The attached report is proposed for publication and is submitted for editing
and subsequent review by the OSI Intelligence Board.
a. The suggested category of publication is: SIR
(SIR, PSIR, SIM, or SIRA)
. All necessary coordination has been achieved as follows:
OSI: Name
Division
Date
Nathe
Division
Date
Name
Division
�Date
CIA: Name
Office
Date
Name
Office
Date
Name
Office
Date
Other: Name
Office
Date
c. Dissemination recommended for the printed report is as follows:
V'USIB Agencies LZNASA
L7PSAC
Nor ad
anada
Australia
New Zealand
(If special dissemination other than that indicated-a-hovels -desired; �
list names and addresses of recipients on the back of this sheet.)
d. The desirable classification for the publication would be
(b)(3)
Approved for Release: 2018/03/28 C06731721
 Approved for Release: 2018/03/28 C06731721
DEPARTMENT OF THE ARMY
UNITED STATES ARMY ENGINEER TOPOGRAPHIC LABORATORIES
FORT BELVO1R, VIRGINIA 22060
.Dear
I would like to
intelligence report:
Title.: Artificial
Author:
September 28, 1983
request a copy of the following scientific
Intelligence Research in the USSR
Date: September 8, 1964
Report No.: OSI-SR/64-37
Monitor Series: AD-F630 175
Please address it to:
U. S. Army Engineer Topographic Laboratories
STINFO Center
Fort Balvoir, VA 22.060
ATM: Mildred Stiger
Thank you.
Sincerely yours,_
2 ,
C '
MN.dred Stiger
Chief, Scientific
'Technical information
Center
��t�) d k
/a--6
keetiud--
At4) Ok Ltr-vt-1
a_4(ta,d-L3-
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
IN REPLY
REFER TO
DEFENSE LOGISTICS AGENCY
DEFENSE DOCUMENTATION CENTER
CAMERON STATION
ALEXANDRIA, VIRGINIA 22314
DTIC-DDAB P83..0383
fiAc),Weliba.0 202-274-6847 or AUTO VON 28-46847)
SUBJECT: Request for Scientific and Technical Reports
TO:
Director
Central Intelligence Agency
Office of Scientific intelligence
HQ Bg
Washington, DC 20505
12 Sep 83
1. We have been unable to locate the report referenced below in the DDC collection, although we believe it to be DoD-funded
and available to us under DoD Instruction 5100.38. We ask that you send us copies in compliance with that instruction.
2. In the event that the instruction.specifies more than two copies for this report, and it would creaks hardship to send the
required number, we, ask that you send only two copies if two copies are unavailable, DDC will accept one copy- or aloan
copy which will be returned in 30 days. All copies must be suitable for photographic reproduction., �
3. UNCLASSIFIED REPORTS MUST BE MARKED WITH A DISTRIBUTION STATEMENT, see DoD Directive 5200.20
(AR 70-31, NAVMATINST 5200.29 or AFR 80-45). Check reverse side, of this ,letter for, proper distribution statement.
4. A franked mailing label for shipping the reports and a DDC Form 50 to obtain the AD number after processing are enclosed.
S. If for any.reason you cannot furnish the report, please return the copy of this letter annotating your reason on the reverse side.
FOR THE ADMINISTRATOR:
2 Encl
I. Franked Label
2. DDC Form 50
HARRY A. SCHREC #f 1ST,
atfeaMealliiitaiel.k9oc
Chief,. Acquisition Section
P83-0383
Central Intelligence Agency Washington, DC
Artificial Intelligence Research in the USSR - by
OSI-SR/64-37 - Scientific Intelligence Rept. . 8 Sep 64
foil
83..255
FL-88
SEP 77
30041
Return this copy to
DDC with reports
or as reply.
Approved for Release: 2018/03/28 C06731721
Approved for Release: 2018/03/28 C06731721
�
DISTRIBUTION STATEMENTS
(DoD Directive 5200.20)
Please check appropriate block:
Cl I. Copies are being forwarded. ,
If the report is unclassified, indicate whether Statement A or B applies.
Statement A.
CI Approved for public release; distribution unlimited.
Statement B.
0 Distribution limited to U.S. Government'agencies only;
(fill in reason); (date statement applied). Other
requests for this document must be referred to (insert
controlling DoD office).
0 2. This document was previously forwarded to DDC on (date)
and the AD number is
3. In accordance with the provisions of Department of Defense
instructions, the document requested is not supplied because:
El It is TOP SECRET.
0 It is excepted in accordance with DoD instructions pertaining
to communications and electronic intelligence.
El It is a registered publication.
CI It is a contract or grant proposal, or an order.
0 Will be published at a later date. (Enter approximate date,
if known.)
12?�;ther (Give reason).
(b)(3)
(b)(6)
Approved for Release: 2018/03/28 006731721
Approved for Release: 2018/03/28 C06731721
8 July 1964
,WP 561/07
Document Nurnbe r(s)
OSI-SR/64-37
Artificial Intelli&nce Research in the USSR
Title
Analyst Div.
LSD
This document should be classified as
with this limitation
None
In addition to the standard USIB and non-USIB distribution, it may be
released to:
UK x N. Z.
Can.
COlvilviEN TS:
CC:
NIS
GIS
Others
Approved for Release: 2018/03/28 C06731721