See All News
Eduard Poghosyan
Position: Head of Cognitive Algorithms and Models Direction

Phone: (+374 60) 62-35-57

  Titles, Degree

Professor, PhD and Doctoral in Mathematical Cybernetics

Doctoral degree in Mathematical Cybernetics, Moscow, Computing Centre of the USSR Academy of Sciences, 1985; Thesis: Adaptation of combinatorial algorithms of classification and control. 
MBA program at American University of Armenia (AUA), 1992-94
PhD. in Mathematical Cybernetics, Moscow, Computing Centre of the USSR Academy of Sciences, 1970; Thesis: Systems of Non-comparable Sets with Minimum Number of Subsets and their Applications to Testing (complete solution of Shroder’s problem)
M.S. in Computer Science, Electro technical State University, St. Petersburg, 1966; Thesis: Multi Layer Neuron Network
  Professional Experience

Leader, Cognitive Modeling Research Direction at the Academy of Sciences of Armenia (from1973)
Teaching and PhD/MSc supervision in the field of Artificial Intelligence at leading universities of Armenia -Yerevan State, Engineering, American, Russian (1970-2014); from 2014, teaching and supervising at the  Academy of Sciences of Armenia
Strategy Management/ Marketing:  learned, taught and modelled at American University of Armenia (1994-98)

Member of Association of Cognitive Modeling (from 2010)
Member of ABSEL and US Marketing Associations (from 1995)
Representative of Armenia:  AI Union of USSR (1970-1992)
  Research Interests

  1. I research human cognizing with the aim of constructing and applying its adequate models, to reveal constraints on cognizing and its origination, approaching to understanding the origination of cellular realities.
  2. Two ongoing projects Constructing Adequate Models of Cognizing  and Constructing Adequate Models of Origination of Cognizing aim to examine viability of models of cognizing and its origination based on combinatorial games.
  3. The projects are based on the following key assumptions and research findings:

    A1. Cognitive systems and means of their construction are various compositions of basic 1-/2- place classifiers.
    Only a few means are sufficient to realize those constructions and compositions. We argue A1 by providing decompositions of ongoing constructive cognitive models to the basic classifiers and interpreting in our models the essentials formulated by Jean Piaget asserting that only a few rules are responsible for the development of our cognizing.

    A2. At present, the highest cognitive power of humans brings them close to the constructive modeling of their own self-reproduction, both biologically and cognitively.
    A2 is based on references to current advances in chemical modeling of biological cells and AI advances in cognitive modeling.

    A3. Information can originate in Nature.
    We come to A3 learning from the research by Juan Parrondo and other physicists in “Thermodynamics of information”, NATURE PHYSICS, VOL 11, FEBRUARY 2015, aimed to reveal the ways in which information can originate in Nature.

    A4. Information and classifying are inseparable from each other.
    A1-4 imply the following corollaries:

    Clr1 from A1, A3, A4: Classifiers, cognitive systems and means of their construction can originate in Nature within the frame of the laws of physics.

    Clr2 from A4: The problem of origination of information can be reduced to the origination of classifiers. And, because origination of classifiers seems more tangible, the studies of origination of information, and therefore, negentropicity, get an additional research dimension.

    Clr3 from A1 and Clr1: Non-cellular, constructive cognizers comparable with the highest human ones can originate in Nature.

    Clr4 from A2 and Clr3: In Nature, non-cellular, constructive cognizers can produce, in a variety of ways, descendant cognizers of comparable effectiveness. It is not excluded that existing cellulars, in fact, represent one of those constructed, evolved cognizers. Cellulars are much more complex than, say computers or satellites, which is why their appearance by chance has almost zero probability. At the same time, there are premises to constructive modeling of highest cognizers that, as it was stated in the assumption A2, are able to constructive modeling of their own self-reproduction both biologically and cognitively. Particularly, able to produce cellular cognizers developing themselves to the present day highest human level.

    Clr5/Clr3: If corollary Clr3 takes place, and if conditions similar to those around us (e.g. in our galaxy) are manifold in the Universe, it can be assumed that powerful cognizers can originate in various regions of the Universe and self-develop to the highest levels allowing them to reproduce themselves in a variety of modes.

  4. Our cognitive models are consistent and complimentary to those in AI, allow to interpret communication, i.e., explaining and understanding of communicatives of languages represented both by IDs of meanings of realities and by the samples of input domains of meanings. The models meet also some of requirements by Andrey Linde in his work “Universe, Life, Consciousness”, Stanford , namely, they are explicitly based on the imprints of their causers, realities, while all constructions eventually are, in fact, the compositions of nominated imprints, which, in turn, are the attributes of classifiers both genomic, like sensors, or gained in lifetimes.
  5. Note that realities in our models of cognizing (see the Introduction) include imprints, the causers of imprints and classifiers/attributes. Namely, they are defined as follows:.... Sensors along with other classifiers inherited and identified by controllers in conjunction with those studied and identified in a lifetime, i.e., revealed, discovered but mostly acquired from cultures of communities, comprise our attributes.The outputs of attributes entail imprints in us that we classify to represent the causers of imprints, particularly those caused by impacts of a causer on our utilities. The imprints, their causers and classifiers/attributes are our realities, while the totalities of realities comprise our Universes. Thus, “our realities “are “…not substituting reality of our feelings by a successfully working theory of an independently existing material world”, which is why, we think, they could be the basis for trying to answer “…What if our perceptions are as real (or maybe, in a certain sense, are even more real) than material objects? “, questioned by Andrey Linde.
  • For modeling the interrelationship of observers/cognizers with the Universe within the frame of our combinatorial game models of Human in the Universe , it will be necessary, at first, to specify the aspects of the Universe induced by your questionnaires to represent and examine them in our models.

  • The above also provides certain premises to try to model highly questioned consciousness, to examine the adequacy of the models as well as to try to answer to the Andrey Linde questions on “Will it not turn out, with the further development of science, that the study of the universe and the study of consciousness will be inseparably linked, and that ultimate progress in the onewill be impossible without progress in the other?” Unfortunately, consciousness has no proper denotative description.For example, Jaquez Pitrat in his book “Consciousness and conscience, in artificial beings: the conscience of a conscious machine”, ISTE, London, UK, 2009” provides 6 ongoing versions of consciousness. If some of their versions were convincingly argued , we will be glad to try to model them, followed by examining the adequacy of the models.
  Selected Publications
  1. E. Pogossian, “Constructing Models of Being by Cognizing”, Academy of Sciences of Armenia, p. 496, Yerevan, 2020.
  2. E. Pogossian, “Adaptation of combinatorial algorithms”, Academy of Sciences of Armenia, p. 293, Yerevan, 1983.
  3. E. Pogossian, “ Systems of Non-comparable Sets with Minimum Number of Subsets and their Applications to Testing”,  PhD thesis in Mathematical Cybernetics, Moscow, Computing Centre of the Academy of Sciences of Russia, 1970 (also in the Journal of Combinatorial Analysis. Moscow State University, v. 4, 16-21 and in the Math. Problems of Cybernetics and Comp. Science, Yerevan, v. XVI, (1986) 148-161).
We are pleased to announce that in the frame of the All4R&D project's second open call 26 courses in 6 categories are offered, as well as 4 innovative practices, free of charge. The courses are tailored-made, based on the assessed needs of students and professionals, and trends in the industry. The courses are co-created and are offered in joint collaboration of academic and business organizations from 6 countries (Armenia, Bosnia and Herzegovina, North Macedonia, Germany, Finland, and Austria).

All courses will be in English and employ innovative teaching methods.
Click here for more information.

Open Call for the participation of Young Researchers and Companies in Research projects

We are pleased to announce that the NPUA/IIAP Cooperative R&D Unit has open calls for researchers, MS/Ph.D students, and companies in two new international research projects:
  • Astronomical Objects Classification
  • Performance Optimization System for Hadoop and Spark Frameworks.
We are inviting all the interested parties to participate in these projects.
For more information please follow the link:

  • 0014, Yerevan, Republic of Armenia, 1, P. Sevak str.
  • Phone: (+37410) 28-20-30
  • E-mail: