ArticlePDF Available

Hybrid Bionic Cognitive Architecture for Artificial General Intelligence Agents

Authors:
Roman Dushkin at National Research Nuclear University MEPhI
Roman Dushkin
Vladimir Stepankov at Smart-Tech Ltd.
Vladimir Stepankov
  • Smart-Tech Ltd.
Download full-text PDF
Read full-text
Download citation

Abstract and Figures

The article describes the author’s proposal on cognitive architecture for the development of a general-level artificial intelligent agent («strong» artificial intelligence). New principles for the development of such an architecture are proposed — a hybrid approach in artificial intelligence and bionics. The architecture diagram of the proposed solution is given and descriptions of possible areas of application are described. Strong artificial intelligence is a technical solution that can solve arbitrary cognitive tasks available to humans (human-level artificial intelligence) and even surpass the capabilities of human intelligence (artificial superintelligence). The fields of application of strong artificial intelligence are limitless — from solving current problems facing the human to completely new problems that are not yet available to human civilization or are still waiting for their discoverer. The novelty of the work lies in the author’s approach to the construction of cognitive architecture, which has absorbed the results of many years of research in the field of artificial intelligence and the results of the analysis of cognitive architectures of other researchers.
Figures - uploaded by Roman Dushkin
Author content
All figure content in this area was uploaded by Roman Dushkin
Content may be subject to copyright.
Content uploaded by Roman Dushkin
Author content
All content in this area was uploaded by Roman Dushkin on Jul 23, 2021
Content may be subject to copyright.
ScienceDirect
Available online at www.sciencedirect.com
Procedia Computer Science 190 (2021) 226–230
1877-0509 © 2021 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 2020 Annual International Conference on Brain-Inspired Cognitive
Architectures for Artificial Intelligence: Eleventh Annual Meeting of the BICA Society
10.1016/j.procs.2021.06.028
10.1016/j.procs.2021.06.028 1877-0509
© 2021 The Authors. Published by Elsevier B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientic committee of the 2020 Annual International Conference on Brain-Inspired
Cognitive Architectures for Articial Intelligence: Eleventh Annual Meeting of the BICA Society
Available online at www.sciencedirect.com
ScienceDirect
Procedia Computer Science 00 (2019) 000–000
www.elsevier.com/locate/procedia
1877-0509 © 2021 The Authors. Published by ELSEVIER B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 2020 Annual International Conference on Brain-Inspired Cognitive
Architectures for Artificial Intelligence: Eleventh Annual Meeting of the BICA Society
2020 Annual International Conference on Brain-Inspired Cognitive Architectures for Artificial
Intelligence: Eleventh Annual Meeting of the BICA Society
Hybrid Bionic Cognitive Architecture for Artificial General
Intelligence Agents
Roman V. Dushkinb, Vladimir Y. Stepankova,*
aNational Research Nuclear University MEPhI, Moscow 115409, Russia
bArtificial Intelligence Agency, Volkonsky 1st Lane, 15, 127473 Moscow, Russia
Abstract
The article describes the author's proposal on cognitive architecture for the development of a general-level artificial intelligent
agent («strong» artificial intelligence). New principles for the development of such an architecture are proposed — a hybrid
approach in artificial intelligence and bionics. The architecture diagram of the proposed solution is given and descriptions of
possible areas of application are described. Strong artificial intelligence is a technical solution that can solve arbitrary cognitive
tasks available to humans (human-level artificial intelligence) and even surpass the capabilities of human intelligence (artificial
superintelligence). The fields of application of strong artificial intelligence are limitless — from solving current problems facing
the human to completely new problems that are not yet available to human civilization or are still waiting for their discoverer. The
novelty of the work lies in the author's approach to the construction of cognitive architecture, which has absorbed the results of
many years of research in the field of artificial intelligence and the results of the analysis of cognitive architectures of other
researchers.
© 2021 The Authors. Published by ELSEVIER B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 2020 Annual International Conference on Brain-Inspired
Cognitive Architectures for Artificial Intelligence: Eleventh Annual Meeting of the BICA Society
Keywords: artificial intelligence; strong artificial intelligence; cognitive architecture; artificial intelligent agent; hybrid artificial intelligence;
bionic approach; machine learning; multisensory integration; goal-setting; explainability.
* Corresponding author. Tel.: +7-916-577-7767.
E-mail address: vstepankov@gmail.com
Available online at www.sciencedirect.com
ScienceDirect
Procedia Computer Science 00 (2019) 000–000
www.elsevier.com/locate/procedia
1877-0509 © 2021 The Authors. Published by ELSEVIER B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 2020 Annual International Conference on Brain-Inspired Cognitive
Architectures for Artificial Intelligence: Eleventh Annual Meeting of the BICA Society
2020 Annual International Conference on Brain-Inspired Cognitive Architectures for Artificial
Intelligence: Eleventh Annual Meeting of the BICA Society
Hybrid Bionic Cognitive Architecture for Artificial General
Intelligence Agents
Roman V. Dushkinb, Vladimir Y. Stepankova,*
aNational Research Nuclear University MEPhI, Moscow 115409, Russia
bArtificial Intelligence Agency, Volkonsky 1st Lane, 15, 127473 Moscow, Russia
Abstract
The article describes the author's proposal on cognitive architecture for the development of a general-level artificial intelligent
agent («strong» artificial intelligence). New principles for the development of such an architecture are proposed — a hybrid
approach in artificial intelligence and bionics. The architecture diagram of the proposed solution is given and descriptions of
possible areas of application are described. Strong artificial intelligence is a technical solution that can solve arbitrary cognitive
tasks available to humans (human-level artificial intelligence) and even surpass the capabilities of human intelligence (artificial
superintelligence). The fields of application of strong artificial intelligence are limitless — from solving current problems facing
the human to completely new problems that are not yet available to human civilization or are still waiting for their discoverer. The
novelty of the work lies in the author's approach to the construction of cognitive architecture, which has absorbed the results of
many years of research in the field of artificial intelligence and the results of the analysis of cognitive architectures of other
researchers.
© 2021 The Authors. Published by ELSEVIER B.V.
This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the 2020 Annual International Conference on Brain-Inspired
Cognitive Architectures for Artificial Intelligence: Eleventh Annual Meeting of the BICA Society
Keywords: artificial intelligence; strong artificial intelligence; cognitive architecture; artificial intelligent agent; hybrid artificial intelligence;
bionic approach; machine learning; multisensory integration; goal-setting; explainability.
* Corresponding author. Tel.: +7-916-577-7767.
E-mail address: vstepankov@gmail.com
2 Author name / Procedia Computer Science 00 (2019) 000–000
1. Introduction
Artificial intelligence as a scientific field is already 7 decades old. But even nowadays there is an acute question —
is it possible to create an artificial cognitive agent with strong learning and self-learning abilities? This question is still
debatable, and there is no clear answer to it. The credo of the authors of this work is that there are no fundamental
obstacles to creating a strong artificial intelligence, since we have a clear example of intelligent agents — ourselves.
In other words, if there is an example of an intelligent agent, why should it be impossible to create other intelligent
agents, perhaps of a different nature?
Scientists and engineers working in the field of strong artificial intelligence are currently focused on so -called
cognitive architectures. This is a general presentation of principles and solutions (possibly in some aspects) that can
somehow solve the problem of building a strong artificial cognitive agent. To date, a large number of cognitive
architectures of strong artificial intelligence have been developed and proposed, most of which are based on the study
of the properties and functionality of the human brain and central nervous system [Yates et al., 2020].
Indeed, the anthropocentric approach provides researchers with a solid foundation for trying to develop a strong
artificial intelligence system [Leshchev, 2011]. Since man seems to be an intellectual being, looking at the principles
of nature will allow us to find the correct direction of research at the first stage, which can later be expanded to new
principles (e. g., planes, helicopters and rockets fly through the air not like birds do). Thus, the principles of bionics
[Lipov, 2010] are quite suitable for the initial design of cognitive architectures for artificial general intelligence agents
(strong artificial intelligence).
The purpose of this work is to present a new cognitive architecture for an artificial general intelligence agent, which
is based on the long-term research in the field of artificial intelligence and, as expected, may well become the basis
for further research in scientific collaborations.
2. Principles of the hybrid paradigm of artificial intelligence
Since the founding of artificial intelligence as an interdisciplinary field of research in 1956, the founding fathers
have launched two paradigms of artificial intelligence — bottom-up or «dirty» and top-down or «clean». The first
focuses on modeling the basic elements of the biological substrate that forms the basis of human intelligence —
neurons, as well as on the study of artificial neural networks, through which it tries to reach intelligence as an emergent
property. The second focuses on modeling pure cognitive processes at a high level, thereby making an attempt to
create intelligence per se [Dushkin & Andronov, 2020].
Some researchers note that combining the approaches and methods of both the bottom-up and top-down paradigms
into a single scheme will solve the problems of both paradigms, leaving only their best sides [Dushkin & Andronov,
2020; Tahmasebi, 2012]. The authors' position on this issue is to combine approaches in a single architecture, which
generally corresponds to how perception, cognition and intelligence work at the logical level in humans. The scheme
of operation of an artificial intelligent agent with a hybrid architecture is based on a cyclic repetition of the process of
perception of information from the environment, its processing by sensory neural networks, decision-making using a
symbolic universal output machine, and transmitting the decision to effectors (actuators) for interacting with the
environment through a motor neural network [Dushkin & Andronov, 2020].
In general, this scheme really allows us to neutralize the negative aspects of both paradigms. However, to create
an artificial general intelligence agent in a hybrid architecture of artificial intelligence, there is a lack of an essential
characteristic that the human mind has — understanding the situation in which the agent finds itself, and understanding
it taking into account the context and personal experience [Dushkin, 2020]. Updating knowledge bases «on the fly»
and the ability to choose from several variants of meaning available both in personal experience and in the knowledge
of all mankind — these are still unsolved problems of the hybrid paradigm in particular and of artificial intelligence
in general. This problem is actual in the related areas of the creation of cybernetic organisms, where the
“communication consequences of the technological expansion of the space of cognition” [Leshchev, 2014] are
significant. There are also other nuances that will be discussed in the next section.
Roman V. Dushkin et al. / Procedia Computer Science 190 (2021) 226–230 227
Available online at www.sciencedirect.com
ScienceDirect
Procedia Computer Science 00 (2019) 000–000
www.elsevier.com/locate/procedia
Available online at www.sciencedirect.com
ScienceDirect
Procedia Computer Science 00 (2019) 000–000
www.elsevier.com/locate/procedia
2 Author name / Procedia Computer Science 00 (2019) 000–000
1. Introduction
Artificial intelligence as a scientific field is already 7 decades old. But even nowadays there is an acute question —
is it possible to create an artificial cognitive agent with strong learning and self-learning abilities? This question is still
debatable, and there is no clear answer to it. The credo of the authors of this work is that there are no fundamental
obstacles to creating a strong artificial intelligence, since we have a clear example of intelligent agents — ourselves.
In other words, if there is an example of an intelligent agent, why should it be impossible to create other intelligent
agents, perhaps of a different nature?
Scientists and engineers working in the field of strong artificial intelligence are currently focused on so -called
cognitive architectures. This is a general presentation of principles and solutions (possibly in some aspects) that can
somehow solve the problem of building a strong artificial cognitive agent. To date, a large number of cognitive
architectures of strong artificial intelligence have been developed and proposed, most of which are based on the study
of the properties and functionality of the human brain and central nervous system [Yates et al., 2020].
Indeed, the anthropocentric approach provides researchers with a solid foundation for trying to develop a strong
artificial intelligence system [Leshchev, 2011]. Since man seems to be an intellectual being, looking at the principles
of nature will allow us to find the correct direction of research at the first stage, which can later be expanded to new
principles (e. g., planes, helicopters and rockets fly through the air not like birds do). Thus, the principles of bionics
[Lipov, 2010] are quite suitable for the initial design of cognitive architectures for artificial general intelligence agents
(strong artificial intelligence).
The purpose of this work is to present a new cognitive architecture for an artificial general intelligence agent, which
is based on the long-term research in the field of artificial intelligence and, as expected, may well become the basis
for further research in scientific collaborations.
2. Principles of the hybrid paradigm of artificial intelligence
Since the founding of artificial intelligence as an interdisciplinary field of research in 1956, the founding fathers
have launched two paradigms of artificial intelligence — bottom-up or «dirty» and top-down or «clean». The first
focuses on modeling the basic elements of the biological substrate that forms the basis of human intelligence —
neurons, as well as on the study of artificial neural networks, through which it tries to reach intelligence as an emergent
property. The second focuses on modeling pure cognitive processes at a high level, thereby making an attempt to
create intelligence per se [Dushkin & Andronov, 2020].
Some researchers note that combining the approaches and methods of both the bottom-up and top-down paradigms
into a single scheme will solve the problems of both paradigms, leaving only their best sides [Dushkin & Andronov,
2020; Tahmasebi, 2012]. The authors' position on this issue is to combine approaches in a single architecture, which
generally corresponds to how perception, cognition and intelligence work at the logical level in humans. The scheme
of operation of an artificial intelligent agent with a hybrid architecture is based on a cyclic repetition of the process of
perception of information from the environment, its processing by sensory neural networks, decision-making using a
symbolic universal output machine, and transmitting the decision to effectors (actuators) for interacting with the
environment through a motor neural network [Dushkin & Andronov, 2020].
In general, this scheme really allows us to neutralize the negative aspects of both paradigms. However, to create
an artificial general intelligence agent in a hybrid architecture of artificial intelligence, there is a lack of an essential
characteristic that the human mind has — understanding the situation in which the agent finds itself, and understanding
it taking into account the context and personal experience [Dushkin, 2020]. Updating knowledge bases «on the fly»
and the ability to choose from several variants of meaning available both in personal experience and in the knowledge
of all mankind — these are still unsolved problems of the hybrid paradigm in particular and of artificial intelligence
in general. This problem is actual in the related areas of the creation of cybernetic organisms, where the
“communication consequences of the technological expansion of the space of cognition” [Leshchev, 2014] are
significant. There are also other nuances that will be discussed in the next section.
228 Roman V. Dushkin et al. / Procedia Computer Science 190 (2021) 226–230
Author name / Procedia Computer Science 00 (2019) 000–000 3
3. Bionic principles suitable for cognitive architecture
Bionics is the application of principles, approaches, and working solutions found in nature to the design and
implementation of technical systems [Bionics, 1993]. Since earlier in this work an anthropocentric position was
proclaimed regarding the study of intelligence in order to attempt to recreate artificial general intelligence in the form
of a technical system, there is no doubt that bionic principles can help in this.
In particular, the following approaches found in living intelligent agents seem highly relevant for use in designing
a new generation of cognitive architecture:
1. Multisensory integration [Kranowitz & Silver, 2006], [Harnad, 1990].
2. Feedback from the decision-making subsystems to the sensors for predicting perception [Shumsky, 2020],
[Zalta, 2014].
3. The personal memory of a cognitive agent [Osipov, 2015].
4. Goal setting [Glazunov, 2011].
5. Internal conflict resolution [Sundas et al., 2020].
Thus, the listed (and in general, incomplete) points represent interesting bionic principles that are reasonable to use
when designing the cognitive architecture of a new generation of artificial intelligent agents to move towards creating
a strong artificial intelligence.
4. The architecture of artificial intelligent agent
To develop a new generation of cognitive architecture, it is necessary to consider examples of such architectures
that were developed earlier to solve various problems both in the field of artificial intelligence in general, and when
trying to develop strong artificial intelligence in particular. Although comparative analysis of cognitive architectures
is of significant research interest in itself, this article will cover only a few relevant cognitive architectures. In
particular, the author's interest is focused on the following examples:
1. Jeff Hawkins suggests a low-level architecture for the basic building blocks of a general-level cognitive agent,
and they correspond to the so-called cortical columns in the human cortex [Hawkins & Blakeslee, 2005].
2. Raymond Kurzweil describes a similar architectural solution [Kurzweil, 2012], which is based on pattern
recognition and prediction, and the image is broadly understood as any situation in which an intelligent agent
may find itself.
3. Sergey Shumsky provides an interesting cognitive architecture for building a robot operating system
[Shumsky, 2020], which simultaneously uses three pillars of modern machine learning — supervised learning,
unsupervised learning, and reinforcement learning, and all these types of machine learning are used
simultaneously to solve various tasks facing an intelligent agent at different levels.
4. Peter Anokhin developed the theory of functional systems [Anokhin, 1975], in which he described the
cognitive architecture of a living intelligent agent.
Using the described hybrid architecture of artificial intelligence, the bionic principles and the listed examples of
cognitive architectures of early researchers, it is proposed to consider a new generation of cognitive architecture, the
General scheme of which is shown in the following figure.
4 Author name / Procedia Computer Science 00 (2019) 000–000
Fig. 1. General scheme of the new generation cognitive architecture.
The general process of cognition, which is defined by the presented cognitive architecture, consists of the following
steps:
1. Various sensor systems of the intelligent agent receive signals from the environment, as well as predictive
signals from the proactive control subsystem. Based on the actual perceived and expected images, recognizable
images are formed and sent to the reactive control subsystem and the multisensory integration center.
2. A reactive (fast) control subsystem causes an instantaneous reflex reaction of an intelligent agent in cases
where the rule of excitation of such a reaction is present in its structure. If there are no response rules, attention
to the perceived situation is escalated to the proactive control subsystem. If a response rule exists, it is executed
through the action result acceptor.
3. The multisensory integration center builds a complete description of the perceived situation in the environment
of an intelligent agent and sends it to a proactive control subsystem.
4. The proactive (slow) management subsystem interacts with the general knowledge base and personal
experience of the intelligent agent, getting the necessary meaning of the perceived situation from them and
recording new knowledge obtained by the agent as a result of receiving information from the environment. In
addition, the proactive control subsystem uses cycles of interaction with the mechanisms of emotional goal-
setting and conflict resolution to make a final decision about the agent's behavior program in the perceived
situation. To do this, we use a dynamic model of the environment and the agent in it, which is part of the most
proactive control subsystem (not shown in the diagram). The final decision is sent to the action result acceptor
for execution, and also transmitted to the reactive control subsystem for subsequent instant reaction of the
agent to similar situations.
5. The action result acceptor receives the decision made and implements it through the actuators of the intelligent
agent that affect the environment. In this case, the acceptor forms a mode of waiting for the result of the
behavioral program execution, which is satisfied when the result is received. In this case, reinforcement
Roman V. Dushkin et al. / Procedia Computer Science 190 (2021) 226–230 229
4 Author name / Procedia Computer Science 00 (2019) 000–000
Fig. 1. General scheme of the new generation cognitive architecture.
The general process of cognition, which is defined by the presented cognitive architecture, consists of the following
steps:
1. Various sensor systems of the intelligent agent receive signals from the environment, as well as predictive
signals from the proactive control subsystem. Based on the actual perceived and expected images, recognizable
images are formed and sent to the reactive control subsystem and the multisensory integration center.
2. A reactive (fast) control subsystem causes an instantaneous reflex reaction of an intelligent agent in cases
where the rule of excitation of such a reaction is present in its structure. If there are no response rules, attention
to the perceived situation is escalated to the proactive control subsystem. If a response rule exists, it is executed
through the action result acceptor.
3. The multisensory integration center builds a complete description of the perceived situation in the environment
of an intelligent agent and sends it to a proactive control subsystem.
4. The proactive (slow) management subsystem interacts with the general knowledge base and personal
experience of the intelligent agent, getting the necessary meaning of the perceived situation from them and
recording new knowledge obtained by the agent as a result of receiving information from the environment. In
addition, the proactive control subsystem uses cycles of interaction with the mechanisms of emotional goal-
setting and conflict resolution to make a final decision about the agent's behavior program in the perceived
situation. To do this, we use a dynamic model of the environment and the agent in it, which is part of the most
proactive control subsystem (not shown in the diagram). The final decision is sent to the action result acceptor
for execution, and also transmitted to the reactive control subsystem for subsequent instant reaction of the
agent to similar situations.
5. The action result acceptor receives the decision made and implements it through the actuators of the intelligent
agent that affect the environment. In this case, the acceptor forms a mode of waiting for the result of the
behavioral program execution, which is satisfied when the result is received. In this case, reinforcement
230 Roman V. Dushkin et al. / Procedia Computer Science 190 (2021) 226–230
Author name / Procedia Computer Science 00 (2019) 000–000 5
learning mechanisms should record the completed program in the agent's personal experience database as
«good».
It should be noted that the described cycle repeats constantly, continuously, and even in a competitive mode with
itself, since the perception of an intelligent agent must be continuous up to some level of sampling. Therefore, in the
described process, it is also affected by the history of performing the same cognitive process in previous time cycles.
5. Conclusion
The article presents a general scheme of the cognitive architecture of a new generation artificial intelligence agent
based on the principles of the hybrid paradigm of artificial intelligence and bionics. This cognitive architecture has
absorbed some of the suggestions of other researchers in the field of artificial general intelligence. The resulting
scheme is generalized and universal, which makes it possible to obtain artificial intelligent agents of various types
when it is concretized by blocks for solving individual cognitive problems.
References
[1] Anokhin P. K. (1975) Essays on the physiology of functional systems. Moscow, 1975.
[2] Bionics: Nature as a Model (1993). — PRO FUTURA Verlag GmbH, München, Um-weltstiftung WWF Deutschland, 1993.
[3] Dushkin R.V. (2020) Criticism of the "Chinese Room" by J. Searle from the perspective of a hybrid model for development of artificial
cognitive agents. Siberian Journal of Philosophy, 2020, Volume 18, No. 2. pp. 30-47. — DOI: 10.25205 / 2541-7517-2020-18-2-30-47.
[4] Dushkin R. V., Andronov M. G. The Hybrid Design for Artificial Intelligence Systems // In book: Arai K., Kapoor S., Bhatia R. (eds)
Proceedings of the 2020 Intelligent Systems Conference (IntelliSys), Volume 1 (1250). — Springer, Cham, 2020. — P. 164-170. — ISBN
978-3-030-55179-7. — DOI: https://doi.org/10.1007/978-3-030-55180-3_13.
[5] Glazunov Yu. T. (2011) Emotional experience in the system of human goal-setting. Vestnik MGTU, volume 14, No. 1, 2011, pp. 126-140.
[6] Harnad S. (1990) The Symbol Grounding Problem. — Physica, 1990. — D 42: p. 335-346. — URL: https://clck.ru/RD7qu (Accessed
04.10.2020).
[7] Hawkins J., Blakeslee S. (2005) On Intelligence. — New York, NY: Owl Books. — ISBN 978-0-8050-7853-4.
[8] Kranowitz C. S., Silver L. B. (2006) The Out-of-Sync Child. — Penguin Books, 2006. — 352 p. — ISBN 978-0-39953-271-9.
[9] Kurzweil R. (2012) How to Create a Mind: The Secret of Human Thought Revealed. — New York: Viking Books. — ISBN 978-0-670-02529-
9.
[10] Leshchev, S.V. (2014) Interfaces of social ecology from technological convergence to the Internet of Things, Filosofskie Nauki, No. 11,
pp.106–113. (In Russ.)
[11] Leshchev, S.V. (2011). Cognitive Phenomenology and Artificial Intelligence: Communication, Neurophysiology, Technology. — Polygnosis.
№ 3-4. 2011. — Ch. 3-4, pp. 16-25. (In Russ.)
[12] Lipov A. N. (2010) At the origins of modern bionics. Bio -morphological formation in an artificial environment. — Polygnosis. № 1-2. 2010.
— Ch. 1-2, pp. 126-136.
[13] Osipov G. S. (2015) Signs-Based vs. Symbolic Models // Advances in Artificial Intelligence and Soft Computing. 2015.
[14] Shumsky S. A. (2020) Machine intelligence. Essays on the theory of machine learning and artificial intelligence. Moscow, RIOR Publ., 2020.
— 340 p. — ISBN: 978-5-369-01832-3.
[15] Sundas A., Bhatia A., Saggi M., Ashta J. (2020) Reinforcement Learning // In book: Machine Learning and Big Data: Concepts, Algorithms,
Tools, and Applications. — John Wiley & sons, July 2020.
[16] Tahmasebi H. (2012) A hybrid neural networks-fuzzy logic-genetic algorithm for grade estimation. — Computers & Geosciences. 42: p. 18-
27. — Bib-code:2012CG.....42...18T. — DOI: 10.1016/j.cageo.2012.02.004.
[17] Yates S., Walker A., Kerri V. M. (2020) Artificial Intelligence, 2020.
[18] Zalta E. N. (2014) Gottlob Frege // Stanford Encyclopedia of Philosophy (Fall 2014), Edward N. Zalta (ed.).
... This endeavor is not merely technical but philosophical, questioning the very nature of intelligence and consciousness 74 . Some researchers have proposed developing AGI by mimicking the human brain's structure and functions, incorporating elements such as sensory processing and memory storage 75,76 . Questions surrounding sentience, consciousness, and moral responsibility become increasingly relevant as AGI systems approach human-like reasoning and decision-making abilities. ...
Navigating artificial general intelligence development: societal, technological, ethical, and brain-inspired pathways
Article
Full-text available
  • Mar 2025
This study examines the imperative to align artificial general intelligence (AGI) development with societal, technological, ethical, and brain-inspired pathways to ensure its responsible integration into human systems. Using the PRISMA framework and BERTopic modeling, it identifies five key pathways shaping AGI’s trajectory: (1) societal integration, addressing AGI’s broader societal impacts, public adoption, and policy considerations; (2) technological advancement, exploring real-world applications, implementation challenges, and scalability; (3) explainability, enhancing transparency, trust, and interpretability in AGI decision-making; (4) cognitive and ethical considerations, linking AGI’s evolving architectures to ethical frameworks, accountability, and societal consequences; and (5) brain-inspired systems, leveraging human neural models to improve AGI’s learning efficiency, adaptability, and reasoning capabilities. This study makes a unique contribution by systematically uncovering underexplored AGI themes, proposing a conceptual framework that connects AI advancements to practical applications, and addressing the multifaceted technical, ethical, and societal challenges of AGI development. The findings call for interdisciplinary collaboration to bridge critical gaps in transparency, governance, and societal alignment while proposing strategies for equitable access, workforce adaptation, and sustainable integration. Additionally, the study highlights emerging research frontiers, such as AGI-consciousness interfaces and collective intelligence systems, offering new pathways to integrate AGI into human-centered applications. By synthesizing insights across disciplines, this study provides a comprehensive roadmap for guiding AGI development in ways that balance technological innovation with ethical and societal responsibilities, advancing societal progress and well-being.
View
... Based on the summary of literatures (Table 1), current research on HMCD mainly falls into three aspects: (1) From the perspective of implementation of intelligent systems, study how to transform human cognitive abilities into machine functions to enhance cognitive intelligence and collaborative decision-making abilities of machines [6,[25][26][27][28][29]. (2) Discussing the mechanism and principle of human-machine intelligence integration ranging from fundamental information fusion technologies [30,31] to highlevel human-machine cognitive integration and mutual trust of humans and machines [32,33]. Human-machine trust mainly focuses on two aspects: the interpretability of [34,35] and consistent cognition of humans and machines [36][37][38]. (3) Human-machine collaborative decision-making process, mainly exploring humanmachine interaction modes in decision-making [14,39,40], such as agency allocation between humans and machines [41,42] and how do humans and machines reach consensus on decision-making [43][44][45], etc. In practice, the cognitive level human-machine collaboration systems are popular with the widely application of robots in various industries. ...
Human-machine Collaborative Decision-making: An Evolutionary Roadmap Based on Cognitive Intelligence
Article
Full-text available
  • Jun 2023
With the development of artificial intelligence technology, intelligent machines are increasingly equipped with human-like abilities such as autonomous decision-making, reasoning, active interaction, and situation awareness. Intelligent machines can act as peers to humans and collaborate with humans to complete decision tasks. The ability to collaborate with humans has become an indicator of the intelligence level of a machine, and determines the scope and depth of its applications. Human-machine collaborative decision-making has attracted attentions from multi-disciplines in recent years, and the diverse origins of its developments make the mechanism of collaborative decision-making ambiguous. A thorough combing of the evolution of human-machine collaboration based on cognitive intelligence is of great importance for understanding the nature of human-machine collaboration at the decision layer and guiding future studies. This article makes a retrospect on the evolution of human-machine collaborative decision-making based on cognition intelligence. It summarizes current research in three categories: the human-machine collaborative system implementation, the human-machine intelligence integrative mechanism and the human-machine interaction in collaborative decision-making process. It reveals the roadmap of the evolution of intelligent machines toward human-machine integration intelligence. Based on the roadmap, prospects for future research of human-machine collaborative decision-making are discussed.
View
... To solve these problems, the authors proposed a new mathematical formalism -associative-heterarchical memory (AH-memory), the structure and functioning of which are based both on bionic principles and on the achievements of both paradigms of artificial intelligence [15]. AH-memory is based on this understanding of the structure of the kognitome [14]. ...
The Structure of Associative Heterarchical Memory
Article
  • Nov 2022
Natural language processing by artificial intelligence (NLP) remains an urgent problem of our time. The main task of NLP is to create programs capable of processing and understanding natural languages. To solve these problems the authors have proposed a new mathematical formalism - the associative-heterarchical memory (AH-memory), which structure and functioning are based both on bionic principles and on the achievements of both paradigms of artificial intelligence. This article will provide a comprehensive description of AH-memory. The article will be of interest to developers of artificial intelligence and specialists in the field of NLP.
View
... To solve these problems, the authors proposed a new mathematical formalism -associative-heterarchical memory (AH-memory), the structure and functioning of which are based both on bionic principles and on the achievements of both paradigms of artificial intelligence [15]. AH-memory is based on this understanding of the structure of the kognitome [14]. ...
Structure of associative heterarchical memory
Article
Full-text available
  • Oct 2022
Objectives . Since the 20th century, artificial intelligence methods can be divided into two paradigms: top-down and bottom-up. While the methods of the ascending paradigm are difficult to interpret as natural language outputs, those applied according to the descending paradigm make it difficult to actualize information. Thus, natural language processing (NLP) by artificial intelligence remains a pressing problem of our time. The main task of NLP is to create applications that can process and understand natural languages. According to the presented approach to the construction of artificial intelligence agents (AI-agents), processing of natural language should be conducted at two levels: at the bottom, methods of the ascending paradigm are employed, while symbolic methods associated with the descending paradigm are used at the top. To solve these problems, the authors of the present paper propose a new mathematical formalism: associative heterarchical memory (AH-memory), whose structure and functionality are based both on bionic principles and on the achievements of top-down and bottom-up artificial intelligence paradigms. Methods . Natural language recognition algorithms were used in conjunction with various artificial intelligence methods. Results . The problem of character binding as applied to AH-memory was explored by the research group in earlier research. Here, abstract symbol binding was performed using multi-serial integration, eventually converting the primary symbols produced by the program into integrated abstract symbols. The present paper provides a comprehensive description of AH-memory in the form of formulas, along with their explanations and corresponding schemes. Conclusions . The most universal structure of AH-memory is presented. When working with AH-memory, a developer should select from a variety of possible module sets those AH-memory components that support the most successful and efficient functioning of the AI-agent.
View
... the implementation of arti cial general intelligence, arti cial intelligence that exceeds the level of human intelligence in some areas is called super arti cial intelligence [11].at present, the drawbacks of special arti cial intelligence are gradually emerging, and arti cial general intelligence has gradually become a new hot spot [12].with the new development of neuroscience[13] and cognitive science14, there have been research on the hardware direction of simulating brain neurons and structures, and the software aspect has generally followed the three major research methods of special arti cial intelligence, such as the NARS system [15][16][17].but progress is still not optimistic, and there are no public reports of the formation of arti cial general intelligence.from the perspectives of philosophy, physics, neuroscience, cognitive science, linguistics, computer science, etc., based on the principles of universal generated philosophy, this paper proposes a new a model of generated arti cial general intelligence, explains its philosophical principles, construction models, implementation models, and attaches examples. ...
Generated Artificial General Intelligence--The Philosophical Principle of Artificial General Intelligence and Give an Example
Preprint
Full-text available
  • Aug 2022
In recent years, with the development of brain science, neuroscience and cognitive science, artificial intelligence technology has made a series of achievements. However, it still fails to achieve the human level of universal artificial intelligence, and the cognitive structure and consciousness are still unsolved mysteries. This paper integrates the evolutionary laws of the universe, life and thinking, summarizes a model of generated general intelligence and reveal its philosophical principle and algorithm structure, then calculates the functions of thinking and consciousness one by one. The results show that the model and its based principles and algorithms conform to the characteristics of biology, physics, neuroscience, cognitive science and philosophy of intelligent species. It is an implementation model of artificial general intelligence that simulates human intelligence. This paper reveals the characteristics of cognition, thinking and consciousness, which also has a good enlightenment to the operation mode of human cognition, thinking and consciousness.
View
... To solve these problems, the authors proposed a new mathematical formalism -associative-heterarchical memory (AH-memory), the structure and functioning of which are based both on bionic principles and on the achievements of both paradigms of artificial intelligence [15]. AH-memory is based on this understanding of the structure of the kognitome [14]. ...
The Structure of Associative Heterarchical Memory
Article
  • Jan 2022
Natural language processing by artificial intelligence (NLP) remains an urgent problem of our time. The main task of NLP is to create programs capable of processing and understanding natural languages. To solve these problems the authors have proposed a new mathematical formalism - the associative-heterarchical memory (AH-memory), which structure and functioning are based both on bionic principles and on the achievements of both paradigms of artificial intelligence. This article will provide a comprehensive description of AH-memory. The article will be of interest to developers of artificial intelligence and specialists in the field of NLP.
View
... Все остальные процессы, происходящие в рамках представленной архитектуры, осуществляются в полном соответствии с той гибридной архитектурой искусственного интеллектуального агента, которая представлена в работе [15]. ...
Principles of Solving the Symbol Grounding Problem in the Development of the General Artificial Cognitive Agents
Article
Full-text available
  • Jul 2022
The article describes the author’s approach to solving the problem of symbol grounding, which can be used in the development of artificial cognitive agents of the general level. When implementing this approach, such agents can receive the function of understanding the sense and context of the situations in which they find themselves. The article gives a brief description of the problem of understanding the meaning and sense. In addition, the author’s vision is given of how the symbol grounding should occur when the artificial cognitive agent uses sensory information flows of various modality. Symbol grounding is carried out by building an associative-heterarchical network of concepts, with the help of which the hybrid architecture of an artificial cognitive agent is expanded. The novelty of the article is based on the author’s approach to solving the problem, which is represented by several important principles — these are multisensory integration, the use of an associative-heterarchical network of concepts and a hybrid paradigm of artificial intelligence. The relevance of the work is based on the fact that today the problem of constructing artificial cognitive agents of a general level is becoming more and more important for solving, including within the framework of national strategies for the development of artificial intelligence in various countries of the world. The article is of a theoretical nature and will be of interest to specialists in the field of artificial intelligence, as well as to all those who want to stay within the framework of modern trends in the field of artificial intelligence.
View
Optimal Scheduling of Processing Unit Using Convolutional Neural Network Architecture
Chapter
Full-text available
  • May 2023
CPU Scheduling is the process of allocating CPU time to various processes of different kinds. There are many existing algorithms that schedule waiting processes, but each of those algorithms achieve good results in only one of the many useful features of a scheduler. Some important features of a scheduling algorithm are to reduce the waiting time, to give a fair share of CPU time to all the processes and to give preference to higher priority processes; Shortest Job First, Round Robin and Priority scheduling algorithms do them respectively. The proposed work combines all these desired properties into one algorithm, by making use of convolution neural network architecture. Using CNN architecture is advantageous because the data is controllable in the hidden layers. The data in the hidden layers could be both understood and manipulated; hence a more powerful neural network could be designed. In comparison to these common algorithms the proposed work achieves 66% better performance, when all the above mentioned desired properties are taken into consideration.KeywordsCPU schedulingOperating SystemNeural networksConvolutional neural networksMin-pooling
View
View
Principles of Solving the Symbol Grounding Problem in the Development of the General Artificial Cognitive Agents
Chapter
  • Sep 2022
The article describes the author’s approach to solving the problem of symbol grounding, which can be used in the development of artificial cognitive agents of the general level. When implementing this approach, such agents can receive the function of understanding the sense and context of the situations in which they find themselves. The article gives a brief description of the problem of understanding the meaning and sense. In addition, the author’s vision is given of how the symbol grounding should occur when the artificial cognitive agent uses sensory information flows of various modality. Symbol grounding is carried out by building an associative-heterarchical network of concepts, with the help of which the hybrid architecture of an artificial cognitive agent is expanded. The novelty of the article is based on the author’s approach to solving the problem, which is represented by several important principles—these are multisensory integration, the use of an associative-heterarchical network of concepts and a hybrid paradigm of artificial intelligence. The relevance of the work is based on the fact that today the problem of constructing artificial cognitive agents of a general level is becoming more and more important for solving, including within the framework of national strategies for the development of artificial intelligence in various countries of the world. The article is of a theoretical nature and will be of interest to specialists in the field of artificial intelligence, as well as to all those who want to stay within the framework of modern trends in the field of artificial intelligence. KeywordsMeaningSenseFrege’s triangleSymbol groundingSemanticsUnderstandingArtificial intelligenceMultisensory integrationAssociative-heterarchical networkHybrid cognitive architecture
View
On J. Searle’s «Chinese Room» from the Hybrid Model of the Artificial Cognitive Agents Design
Article
Full-text available
  • Jan 2020
The article presents a review of the phenomenon of understanding the meaning of the natural language and, more broadly, the meaning of the situation in which the cognitive agent is located, considering the context. A specific definition of understanding is given, which is at the intersection of neurophysiology, information theory and cybernetics. The scheme of an abstract architecture of the cognitive agent (of arbitrary nature) is offered, which states that an agent with such architecture can understand in the sense described in the paper. It also provides a critique of J. Searle’s mental experiment “The Chinese Room” from the point of view of the construction of artificial cognitive agents within a hybrid paradigm of artificial intelligence. The novelty of the presented work is based on the application of the author’s methodological approach to the construction of artificial cognitive agents. It not only considers the perception of external stimuli from the environment, but also the philosophical problem of “understanding” by the artificial cognitive agent of its sensory inputs. The relevance of the work follows from the renewed interest of the scientific community in the theme of Strong Artificial Intelligence (or AGI). The author's contribution consists in comprehensive treatment from different points of view of the theme of understanding perceived by artificial cognitive agents. It involves the formation of prerequisites for the development of new models and the theory of understanding within the framework of artificial intelligence, which in the future will help to build a holistic theory of the nature of human mind. The article will be interesting for specialists working in the field of artificial intellectual systems and cognitive agents construction, as well as for scientists from other scientific fields – first of all, philosophy, neurophysiology and psychology.
View
The Hybrid Design for Artificial Intelligence Systems
Chapter
Full-text available
  • Jan 2021
The article discusses approaches to intelligent systems building based on a hybrid paradigm implemented by combining the bottom-up (neural network) and top-down (symbolic) approaches to the design and development of artificial intelligence systems. The scheme of the hybrid intelligence system device is described, its architecture, the purpose and functionality of its components, the principles of operation and the use of its subsystems are explained.
View
A hybrid neural networks-fuzzy logic-genetic algorithm for grade estimation
Article
Full-text available
  • May 2012
  • COMPUT GEOSCI-UK
The grade estimation is a quite important and money/time-consuming stage in a mine project, which is considered as a challenge for the geologists and mining engineers due to the structural complexities in mineral ore deposits. To overcome this problem, several artificial intelligence techniques such as Artificial Neural Networks (ANN) and Fuzzy Logic (FL) have recently been employed with various architectures and properties. However, due to the constraints of both methods, they yield the desired results only under the specific circumstances. As an example, one major problem in FL is the difficulty of constructing the membership functions (MFs).Other problems such as architecture and local minima could also be located in ANN designing. Therefore, a new methodology is presented in this paper for grade estimation. This method which is based on ANN and FL is called “Coactive Neuro-Fuzzy Inference System” (CANFIS) which combines two approaches, ANN and FL. The combination of these two artificial intelligence approaches is achieved via the verbal and numerical power of intelligent systems. To improve the performance of this system, a Genetic Algorithm (GA) – as a well-known technique to solve the complex optimization problems – is also employed to optimize the network parameters including learning rate, momentum of the network and the number of MFs for each input. A comparison of these techniques (ANN, Adaptive Neuro-Fuzzy Inference System or ANFIS) with this new method (CANFIS–GA) is also carried out through a case study in Sungun copper deposit, located in East-Azerbaijan, Iran. The results show that CANFIS–GA could be a faster and more accurate alternative to the existing time-consuming methodologies for ore grade estimation and that is, therefore, suggested to be applied for grade estimation in similar problems.
View
Artificial intelligence and machine learning
Article
  • Sep 2021
Artificial Intelligence (AI) is an area of research driven by innovation and development that culminates in computers, machines with human-like intelligence characterized by cognitive ability, learnability, adaptability and decision-making ability. The study found that AI is widely adopted and used in education, especially by educational institutions, in various forms. This article reviewed articles by various scientists from different countries. The paper discusses the prospects for the application of artificial intelligence and machine learning technologies in education and in everyday life. The history of the development of artificial intelligence is described, technologies of machine learning and neural networks are analyzed. An overview of already implemented projects for the use of artificial intelligence is given, a forecast of the most promising, according to the authors, directions for the development of artificial intelligence technologies for the next period is given. This article provides an analysis of how educational research is being transformed into an experimental science. AI is combined with the study of science into new ‘digital laboratories’, in which ownership of data, as well as power and authority in the production of educational knowledge, are redistributed between research complexes of computers and scientific knowledge.
View
Signs-Based vs. Symbolic Models
Conference Paper
  • Oct 2015
  • Lect Notes Comput Sci
In this paper a sign-based or semiotic formalism is considered. The concept of sign arose in the framework of semiotics. Neurophysiological and psychological researches indicate sign-based structures, which are the basic elements of the world model of a human subject. These elements are formed during his/her activity and communication. In this formalism it was possible to formulate and solve the problem of goal-setting, i.e. generating the goal of behavior.
View
View
The Symbol Grounding Problem
Article
  • Jun 1990
  • PHYSICA D
How can the semantic interpretation of a formal symbol system be made intrinsic to the system, rather than just parasitic on the meanings in our heads? How can the meanings of the meaningless symbol tokens, manipulated solely on the basis of their (arbitrary) shapes, be grounded in anything but other meaningless symbols? The problem is analogous to trying to learn Chinese from a Chinese/Chinese dictionary alone. A candidate solution is sketched: Symbolic representations must be grounded bottom-up in nonsymbolic representations of two kinds: (1) "iconic representations," which are analogs of the proximal sensory projections of distal objects and events, and (2) "categorical representations," which are learned and innate feature-detectors that pick out the invariant features of object and event categories from their sensory projections. Elementary symbols are the names of these object and event categories, assigned on the basis of their (nonsymbolic) categorical representations. Higher-order (3) "symbolic representations," grounded in these elementary symbols, consist of symbol strings describing category membership relations (e.g., "An X is a Y that is Z").
View
View
Essays on the physiology of functional systems
  • Jan 1975
  • P K Anokhin
Anokhin P. K. (1975) Essays on the physiology of functional systems. Moscow, 1975.
The Out-of-Sync Child. -Penguin Books
  • Jan 2006
  • C S Kranowitz
  • L B Silver
Kranowitz C. S., Silver L. B. (2006) The Out-of-Sync Child. -Penguin Books, 2006. -352 p. -ISBN 978-0-39953-271-9.