The theory of the sign picture of the world
One of the promising directions in AI is the approach to constructing a general model of the human picture of the world and an algorithmic description on its basis of key cognitive functions of planning behavior, goal-setting, and some types of reasoning.
The approach is based on the cultural-historical approach of Vygotsky and Leontiev's theory of activity. The main active element of the picture of the world suggests using the concept of a sign. The sign, as is customary in the activity approach, has four components: image (sensory tissue), meaning, personal meaning and name. The worldview model is based on the interaction of four types of semantic networks, in which the vertices are the components of signs. Cognitive processes in the model are realized due to the spread of activity, the mechanisms of which are different for each type of semantic network on the components of the sign.
Выступление на конференции AIJourney-2019
А.И. Панов на тему "Cognitive Dynamics Systems and Articial General Intelligence" (1:31-1:58)
An elementary stage in the work of each cognitive function is to stop the propagation of activity and link the active nodes of each of the networks into a sign, which, depending on the task, is a target sign, a situation sign or a role sign. Thus, in the model of a sign picture of the world, the cognitive process is a sequence of activation (or formation) of signs.
A significant difference between the proposed mechanism for the propagation of activity from those existing in works on artificial intelligence is the interaction of four types of networks, in contrast to the "single-network" models of neural networks, semantic networks and Petri nets.
As part of this area, we develop not only theoretical methods and new algorithms, but also use them in practice to solve applied problems: visual question answering, planning the behavior of mobile robots and unmanned vehicles, etc.
Publications:
  • Osipov G.S., Panov A.I., Chudova N.V. Behavior control as a function of consciousness. I. Picture of the world and goal-setting // Bulletin of the Russian Academy of Sciences. Theory and control systems. 2014. No. 4. P. 49–62. РИНЦ
  • Osipov G.S., Panov A.I., Chudova N.V. Behavior control as a function of consciousness. II. Synthesis of a plan of behavior // News of the Russian Academy of Sciences. Theory and control systems. 2015. No. 6. P. 47–61. РИНЦ
  • Makarov D.A., Panov A.I., Yakovlev K.S. Architecture of a multilevel intelligent control system for unmanned aerial vehicles // Artificial Intelligence and Decision Making. 2015. No. 3. P. 18–33. РИНЦ
  • Emel'yanov S. et al. Multilayer cognitive architecture for UAV control // Cognitive Systems Research. 2016. Vol. 39. P. 58–72. Elsiever
  • Panov A.I., Yakovlev K.S. Psychologically Inspired Planning Method for Smart Relocation Task // Procedia Computer Science. Elsevier, 2016. Vol. 88. P. 115-124. ScienceDirect
  • Panov A.I., Yakovlev K.S. Interaction of strategic and tactical planning of the behavior of coalitions of agents in a dynamic environment // Artificial Intelligence and Decision Making. 2016. No. 4. P. 68–78. ИИПР
  • Panov A.I., Yakovlev K. Behavior and Path Planning for the Coalition of Cognitive Robots in Smart Relocation Tasks // Robot Intelligence Technology and Applications 4. Advances in Intelligent Systems and Computing / ed. Kim J.-H. et al. Springer, 2017. Vol. 447. P. 3–20. Springer
  • Kiselev G.A., Panov A.I. Synthesis of the Behavior Plan for Group of Robots with Sign Based World Model // Interactive Collaborative Robotics. ICR 2017. Lecture Notes in Computer Science / ed. Ronzhin A., Rigoll G., Meshcheryakov R. Springer, 2017. Vol. 10459. P. 83–94. Springer
  • Osipov G.S., Panov A.I. Relations and operations in the symbolic picture of the world of the subject of behavior // Artificial Intelligence and Decision Making. 2017. No. 4. P. 5–22. ИИПР
  • Panov A.I. Behavior Planning of Intelligent Agent with Sign World Model // Biologically Inspired Cognitive Architectures. 2017. Vol. 19.P. 21–31. ScienceDirect
  • Osipov G.S. and other Significant picture of the world of the subject of behavior. Moscow: Fizmatlit, 2018.264 p.
  • Kiselev G., Kovalev A., Panov A.I. Spatial reasoning and planning in sign-based world model // Artificial Intelligence. RCAI 2018. Communications in Computer and Information Science / ed. Kuznetsov S., Osipov G.S., Stefanuk V. Springer, 2018. Vol. 934. P. 1-10. Springer
  • Kiselev G.A., Panov A.I. A sign approach to the problem of distribution of roles in a coalition of cognitive agents // Proceedings of SPIIRAS. 2018.Vol. 2.No. 57, pp. 161–187. СПИИРАН
  • A.I. Panov Goal Setting and Synthesis of a Behavior Plan by a Cognitive Agent // Artificial Intelligence and Decision Making. 2018. No. 2. P. 21–35. РИНЦ
  • Panov A.I. Formation of the figurative component of knowledge of a cognitive agent with a symbolic picture of the world // Information technologies and computing systems. 2018. No. 4. P. 84–96. RSCI
  • Smirnov I.V. et al. Personal cognitive assistant: concept and principles of work // Informatics and its application. 2019.Vol. 13.No. 3.P. 105–113. MathNet
  • Kovalev A.K., Panov A.I. Mental Actions and Modelling of Reasoning in Semiotic Approach to AGI // Artificial General Intelligence. AGI 2019. Lecture Notes in Computer Science / ed. Hammer P. et al. Springer, 2019. Vol. 11654. P. 121–131. Springer
  • Kovalev A.K., Panov A.I., Osipov E. Hyperdimensional Representations in Semiotic Approach to AGI // Artificial General Intelligence. AGI 2020. Lecture Notes in Computer Science. Springer, 2020. Vol. 12177. P. 231–241. Springer
Presentations
  • General Artificial Intelligence - AIJourney-2019. Slides
  • Cognitive Dynamic Systems - КИИ-2018. Slides
  • Performance at AI@MIPT 2018. Slides
Required skills for trainees
  • Excellent knowledge of Python
  • Technical English
  • Knowledge of graph and network theory
  • Ability to formulate formal statements
  • Interest in psychology is encouraged
Research projects topics
  • Reflection and Reflexive Behavior Models
  • Modeling goal-setting processes
  • Approximation models of reasoning in the picture of the world
  • Algorithms for spreading activity over semantic networks
  • Algorithms for generalization and formation of scenarios in the picture of the world
  • Vector symbolic architectures
Key words
Picture of the world, sign, psychology, cognitive functions, reflection, goal-setting, the problem of binding symbols, general artificial intelligence