University of Calgary
UofC Navigation

Main Research Fields of ICIC

The contemporary wonder of sciences and engineering has recently refocused on the starting point of them: how does the brain process internal and external information autonomously rather than imperatively as those of conventional computers? The key research fields of ICIC encompass cognitive informatics, cognitive computing, cognitive computers, abstract intelligence, denotational mathematics, computational intelligence, cognitive linguistics, cognitive systems, cognitive robots, cognitive agents, neuroinformatics, brain informatics and software science.

 

Cognitive Informatics

Definition 1. Cognitive Informatics (CI) is a discipline across computer science, information science, cognitive science, brain science, intelligence science, knowledge science and cognitive linguistics, which investigates into the internal information processing mechanisms and processes of the brain, the underlying abstract intelligence theories and denotational mathematics, and their engineering applications in cognitive computing and computational intelligence.

The theories of informatics and their perceptions on the object of information have evolved from the classic information theory, computational informatics, to cognitive informatics in the last six decades. Informatics, the science of information, can be classified into three generations. The first generation of classic informatics is signal oriented that deals with the communication properties of information [Shannon, 1948; Bell, 1953; Goldman, 1953]. The second generation of computational informatics is data oriented that studies information as properties of the natural world which can be distinctly elicited, generally abstracted, binary represented, and reducibly operational [Turing, 1950]. The first- and second-generation informatics put emphases on external information processing, which are yet to be extended to observe the fundamental fact that human brains are the original sources and final destinations of information. Any information must be cognized by human beings before it is understood, comprehended, and consumed. The contemporary theory of information reveals that information is the third essence of the natural world supplementing to matter and energy. Therefore, the third generation of cognitive informatics is knowledge and intelligence oriented [Wang, 2002, 2003; Wang et al., 2006, 2009], which studies the properties of cognitive information, the mechanisms of abstract intelligence, and underpinning denotational mathematics.

CI is a cutting-edge field, initiated by Prof. Yingxu Wang and his colleagues since 2002, that tackles the fundamental problems shared by computational intelligence, modern informatics, computer science, AI, cybernetics, cognitive science, neuropsychology, medical science, philosophy, cognitive linguistics, and life science. The development and the cross fertilization among the aforementioned science and engineering disciplines have led to an entire range of extremely interesting new research fields known as CI, which studies the internal information processing mechanisms and processes of the natural intelligence – human brains and minds – and the development of the next generation of cognitive computers. CI forges links between a number of natural science and life science disciplines with informatics and computing science [Wang et al, 2003, 2006, 2009, 2010, 2011, 2012].

 

Cognitive Computing

Definition 2. Cognitive Computing (CC) is a novel paradigm of intelligent computing theories and methodologies based on CI that implements computational intelligence by autonomous inferences and perceptions mimicking the mechanisms of the brain.

CC is emerged and developed based on the multidisciplinary basic research in CI. As in his keynote speech “Towards the Next Generation of Cognitive Computers: Knowledge vs. Data Processors” at ICCSA’12, Dr. Yingxu Wang reveals that: “In celebrating the 100th anniversary of Turing’s pioneer work, curiosity may lead to a fundamental question if more intelligent computers that think, perceive and learn may be developed. The Turing and von Neumann machines are generic data processors created on a basic assumption that objects and behavior of any computing problem can be reduced onto the bit level. However, there is an entire range of complex problems in the real world that may impossibly, or at least, inefficiently be reduced onto bits. This is in accordance with the findings in denotational mathematics that most of the complex entities and problems in the real world cannot be abstracted and represented by pure numbers in B (bits) or R (real numbers). The complex objects beyond B and C are the hyper structures (HS) [Wang, 2008, 2009], which is a type of mathematical entities modeled by complex tuples with multiple fields of attributes and constraints as well as their intricate relations. Examples of the complex objects in HS are, inter alia, abstract concepts, complex relations, perceptual information, formal knowledge, intelligent behaviors, behavioral processes, rational decisions, language/visual semantics, causations and generic systems.

“Is it possible to advance the classic computing theories and capabilities closer to those of human brains as a natural knowledge processor that does not reason in B? Instead of reducing every computing problem and solution onto B as in conventional data computers, the next generation of knowledge computers known as cognitive computers need to be able to directly process human knowledge in HS. Because the basic unit of knowledge is an abstract concept in HS, the mathematical model of knowledge is a Cartesian product of power sets of formal concepts. It is recognized that the basic computing behaviors of the CPU of data computers are arithmetic and logical operations in B.and R. However, those of the CPU for cognitive computers are rigorously represented knowledge, causations and semantics in HS and their formal manipulations using denotational mathematics. The mathematical foundations of classic data computers are Boolean algebra and its logical counterparts in B. However, those of the cognitive computers are based on contemporary denotational mathematics [Wang, 2008] such as concept algebra, inference algebra, semantic algebra and process algebra in HS for rigorously modeling and manipulating knowledge, perception, leaning and inferences.

 

Cognitive Computers

Modern computers can be classified into the categories of imperative, autonomic, and cognitive computers from the bottom up. The imperative computers are a passive system based on stored-program controlled behaviors for data processing [von Neumann, 1946, 1958]. The autonomic computers are goal-driven and self-decision-driven machines that do not rely on instructive and procedural information [Kephart and Chess, 2003; IBM, 2006; Wang, 2004, 2007]. Cognitive computers are more intelligent computers beyond the imperative and autonomic computers, which embodies major natural intelligence behaviors of the brain such as thinking, inference, and learning [Wang, 2006, 2007].

Definition 3. A cognitive computer (CogC) ) is an intelligent computer for knowledge and intelligence processing as that of a classic computer for data processing, which is modeled by a parallel architecture as follows:

CogCs are designed to embody machinable intelligence such as computational inferences, causal analyses, knowledge manipulation, machine learning and autonomous problem solving [Wang, 2009]. It is noteworthy that CogC is not centered by a conventional CPU for data manipulations as that of the classic computers with the von Neumann architecture. However, CogC is centered by the four concurrent engines for cognitive knowledge processing such as the inference engine, learn engine, perception engine, and knowledge manipulation engine based on the Layered reference Model of the Brain (LRMB) [Wang et al., 2006]. In the architecture of CogC, AIE embodies autonomous causal inference and reasoning based on inference algebra [Wang, 2011]; CLE drives machine learning of general knowledge expressed in natural languages based on concept algebra [Wang, 2008]; SPE handles sensory information and attentions, as well as their interpretation, in order to establish machine consciousness about the internal status and external environment [Wang et al., 2006]; and KME implements knowledge representation and manipulation based on concept algebra and semantic algebra [Wang, 2012].

Recent studies in cognitive computing reveal that the computing power in computational intelligence can be classified at four levels: data, information, knowledge, and intelligence from the bottom up. Classic computers are designed for imperative data and information processing by stored-program-controlled mechanisms. However, the increasing demand for advanced computing technologies for knowledge and intelligence processing in the high-tech industry and everyday lives requires novel cognitive computers for providing autonomous computing power for various cognitive systems mimicking the natural intelligence of the brain. Cognitive computers will establish a fundamental platform to implement all facets of computational intelligence such as the perceptive, cognitive, instructive, and reflective intelligence.

 

Abstract Intelligence

Intelligence is a driving force or an ability to acquire and use knowledge and skills, or to inference in problem solving. Intelligence plays an irreplaceable role in the transformation between information, matter, and energy according to the Information-Matter-Energy-Intelligence (IME-I) model [Wang, 2009]. It is a profound human wonder on how conscious intelligence is generated as a highly complex cognitive state in human mind on the basis of biological and physiological structures. How natural intelligence functions logically and physiologically? How natural and artificial intelligence are converged on the bases of brain, mathematics, software and intelligence sciences? It was conventionally deemed that only the mankind and advanced species possess intelligence. However, the development of computers, robots, software agents, and autonomous systems indicates that intelligence may also be created or embodied by machines and man-made systems. Therefore, it is one of the key objectives in cognitive informatics and intelligence science to seek a coherent theory for explaining the mechanisms of both natural and artificial intelligence.

Definition 4. Abstract Intelligence (αI) ) is the general mathematical form of intelligence as a complex natural mechanism that transfers information into behaviors and knowledge at the embodied neural, cognitive, functional and logical levels from the bottom up.
αI, in the narrow sense, is a human or a system ability that transforms information into behaviors. While, in the broad sense, αI is any human or system ability that autonomously transfers the forms of abstract information between data, information, knowledge, and behaviors in the brain or cognitive systems. The field of αI studies the foundations of intelligence science focusing the core properties of intelligence as a natural mechanism that transfers information into behaviors and knowledge. The paradigms of ?I are such as natural, artificial, machinable and computational intelligence. The studies in CI and ?I lay a theoretical foundation toward revealing the basic mechanisms of different forms of intelligence. As a result, cognitive computers may be developed, which are characterized as knowledge processors beyond those of data processors in classic computing theories and systems.

 

Denotational Mathematics

It is recognized that the maturity of a scientific discipline is characterized by the maturity of its mathematical (meta-methodological) means. New forms of mathematics are sought, collectively known as denotational mathematics, in order to deal with complex mathematical entities, such as abstract objects, complex relations, perceptual information, abstract concepts, knowledge, intelligent behaviors, behavioral processes and abstract systems, emerged in cognitive informatics, computational intelligence, software science and knowledge science.

Definition 5. Denotational mathematics (DM) is a category of abstract mathematical structures that deals with high-level mathematical entities in hyper-structures (HS) beyond pure numbers in ¡ and their relational and compositional operations.

The history of sciences and engineering shows that many branches of mathematics have been created in order to meet their abstract, rigorous and expressive needs. These phenomena may be conceived as that new types of problems require new forms of mathematics [Pavel, 1993; Wang, 2006]. Therefore, the entire computing theory, as Lewis and Papadimitriou perceived, is about mathematical models of computers and algorithms [Lewis and Papadimitriou, 1998]. Hence, the essences of cognitive informatics and computational intelligence are about denotational mathematical models and means for abstract, natural and machine intelligence.

Extensions of conventional analytic mathematical entities to more complicated ones beyond numbers and sets lead to the contemporary DM [Wang, 2008]. Typical paradigms of DM are such as fuzzy logic [Zadeh, 1965, 2010, 2011], concept algebra [Wang, 2008], inference algebra [Wang, 2011], semantic algebra [Wang, 2012], system algebra [Wang, 2008; Wang, Zadeh and Yao, 2009], real-time process algebra (RTPA) [Wang, 2002, 2007, 2008], granular algebra [Wang, 2012], visual semantic algebra (VSA) [Wang, 2009]. In the family of DM, concept algebra is a DM for rigorous knowledge manipulation and machine learning. Inference algebra is a DM for formal causation manipulation and machine reasoning. Semantic algebra is a DM for formal semantic manipulation and machine comprehension. RTPA is a DM for rigorous architectural and behavior modeling and computational implementation. System algebra is a DM for abstract system modeling and manipulation. Granular algebra is a DM for rigorous model computing granules and their behavioral manipulations. VAS is a DM for formal representation and manipulation of images and visual patterns for cognitive robots and cognitive systems.

DM is not only a powerful mathematical means for rigorously modeling human cognitive processes, but also an efficient methodology for conveying human cognitive abilities into cognitive computers, cognitive robots and cognitive systems where natural languages and classic analytic mathematics did not work or at least inefficient. A wide range of applications of denotational mathematics have been identified in the fields of cognitive informatics, cognitive computing and computational intelligence as described in other sections of this webpage.

 

Computational Intelligence

Definition 6. Computational intelligence (CoI) is an embodying form of abstract intelligence (αI) that implements intelligent mechanisms and behaviors by computational methodologies and software systems, such as expert systems, fuzzy systems, autonomous computing, intelligent agent systems, genetic/evolutionary systems, and autonomous learning systems.

Computational intelligence (programmed) is a paradigm of means and technologies to embody abstract intelligence supplement to natural intelligence (physiological organs), artificial intelligence (man-made), machinable intelligence (wired) and cognitive systems (hybrid).

 

Cognitive Linguistics

Definition 7. Cognitive linguistics (CL) is an emerging discipline that studies the cognitive properties of natural languages and the cognitive models of languages in cognitive computing and computational intelligence.

Studies on formal syntaxes and semantics are initiated in linguistics and natural language processing, which can be traced back to the works of Alfred Taski (1944), Noam Chomski (1956) and Richard Montague (1970). CL is proposed by Gibbs et al. since 1996, which attempts to explain the cognitive processes of language and knowledge acquisition, storage, production and comprehension [Gibbs, 1996; Taylor, 2002; Langlotz, 2006; Evans & Green, 2006; Wang and Berwick, 2012].

Basic studies in CL [Wang, 2009; Wang and Berwick, 2012] include the cognitive linguistic framework, the deductive semantics of languages, deductive grammar of English, cognitive translators, and the cognitive complexity of online text comprehension. The deductive grammar (DG) is an abstract grammar that formally denotes the syntactic rules of a language based on which as a generic formula, valid language sentences can be deductively derived. The Deductive Grammar of English (DGE) can be formally described by EBNF or more efficiently described by RTPA. DEG specifies the formal structures of sentence, clauses, phrases, modifiers, and terminals (lexes), as well as their syntactical rules, in a top-down hierarchy. Based on the formal DGE and supported by a lexical database for all terminals words, a parser can be implemented to autonomously process texts expressed in English according to the formal grammar. The DGE provides not only an essential set of syntactic rules for implementing an English parser for natural language processing, but also the establishment of the general pattern of English sentences based on it any grammatically corrected sentence is an instance of the general pattern. The DGE theory indicates that, instead of analyzing variously infinite instances of sentences in a language via statistical linguistics, a more efficient way is to rigorously specify the general pattern of the language where all sentences fit [Wang and Berwick, 2012]. An interesting finding in comparative language analyses is that the syntaxes and semantics of languages are relative and interchangeable. It shows that the grammar of natural languages is simpler than that of artificial languages such as a programming language. However, the semantics of the former is much more complicated than those of the latter [Wang, 2009].

 

Cognitive Machine Learning

Learning is a cognitive process of knowledge and behavior acquisition. Learning is a complex cognitive process that gains knowledge of something or acquires skills in some actions and practices. The physiological foundation for learning is memory, particularly Long-Term Memory (LTM), where all forms of artifacts in learning such as the objects, results, and the context are represented in neural concept networks. The most significant result of learning is the change of the cognitive model in LTM. Learning also results in behavioral or capability changes in the Action Buffer Memory (ABM), while some of them may not be explicitly observed in a shorter term.

Definition 8. Cognitive learning is a knowledge or behavioral acquisition process of the brain that composes a new concept into the Object-Attribute-Relation (OAR) model of existing knowledge in LTM or creates a new behavior in ABM.

Learning is a creative process that always results in the generation of new knowledge or skills, and the updating of the entire OAR model of a person in LTM. A various forms of learning have been identified in psychology such as the classic conditioning learning, supervised learning, latent learning, and social learning on the basis of behaviorism and associationism [Pavlov, 1928; Smith, 1993; Leahey, 1997; Sternberg, 1998]. In cognitive science, learning is deemed as a relatively permanent change in the behavior, thought, feelings, and knowledge as a consequence of prior experience [Driscoll, 1991; Smith, 1993; Gray, 1994; Pinel, 1997; Reisberg, 2001; Wilson and Frank, 1999].

In cognitive informatics and cognitive computing, the taxonomy of learning is explored that classifies learning into 6 categories known as objective, transitive, behavioral, abstractive, inferential, and complex learning with 38 specific learning forms as summarized in [Wang, 2006]. Based on this work, a theoretical framework of learning is developed and the cognitive processes of autonomous machine learning is elaborated [Wang, 2012]. A Cognitive Learn Engine (CLE) for cognitive computing and machine-based knowledge manipulation is developed [Wang et al., 2011; Tian et al., 2011] on the basis of concept algebra and RTPA.

 

Cognitive Systems

Definition 9. Cognitive systems (CS) are applications of the fundamental theories and generic technologies of cognitive informatics and cognitive computing in a wide range of fields such as, inter alia, cognitive computers, brain science, AI, computational intelligence, computational linguistics, automatic control systems, cognitive robotics, cognitive agents, cognitive Internet, communications, knowledge systems, medical systems and space systems.

 

Cognitive Robots and Cognitive Agents

Definition 10. A cognitive robot (CR) is an autonomous robot that is capable of inference, perception, and learning based on the three-level computational intelligence known as the imperative, autonomic, and cognitive intelligence.

The studies on cognitive robots are rooted in the essences of artificial intelligence [McCarthy et al., 1955; McCulloch, 1943, 1965], cognitive psychology [Newell, 1990; Sternberg, 1997; Anderson and Rosenfeld, 1998] and computational intelligence [Poole et al., 1997]. Fundamental problems in CR studies are what the necessary and sufficient intelligent behaviors of cognitive robots are and what distinguish the intelligent capabilities of cognitive robots from their imperative counterparts. CRs [Wang, 2010] are built on the three levels of the imperative, autonomic, and cognitive intelligence from the bottom up. The representation and modeling of cognitive robots can be carried out by their architectures and behaviors. The former are a framework of a cognitive robot that represents the overall structure, components and interrelations; while the latter are a set of functions and interactions with the architecture of the cognitive robots.

A Reference Model of Cognitive Robots (RMCR) is developed that explains the architectural differences and behavioral characteristics of cognitive robots beyond conventional imperative robots and computing systems [Wang, 2010]. RMCR explains that a cognitive robot can be modeled and implemented by seven forms of intelligent behaviors. The RMCR model reveals that the relationships of the imperative, autonomic, and cognitive behaviors of cognitive robots are hierarchical and inclusive, where any lower layer behavior of a cognitive robot is a subset of those of a higher layer. In other words, any higher layer behavior is a natural extension of those of lower layers. Therefore, the necessary and sufficient conditions of a cognitive robot are the possession of all seven types of behaviors at the three layers according to the RMCR reference model.

Definition 11. A cognitive agent (CA) is an autonomous software agent that possesses cognitive computing and autonomous decision making abilities as well as interactive communication capability to peers, humans and the environment [Wang, 2009].

It is recognized that the functions of CAs are perception-driven and inference-driven behaviors beyond classic imperative and autonomic ones as implemented by conventional imperative computing technologies. The studies on CA and CR may be converged on the same theoretical foundation of cognitive informatics and cognitive computing.

 

Software Science

Definition 12. Software Science (SS) is a discipline of enquiries that studies the theoretical framework of software as instructive and behavioral information, which can be embodied and executed by generic computers in order to create expected system behaviors and machine intelligence.

The latest developments in computer science, theoretical software engineering, cognitive science, cognitive informatics, and intelligence science, and the crystallization of accumulated knowledge by the fertilization of the aforementioned fields, have led to the emergence of a transdisciplinary and convergence field known as SS. The architecture of SS possesses four categories of studies, i.e., theories and methodologies, denotational mathematics, cognitive informatics and organizational theories.

SS investigates into the common objects in the abstract world such as software, information, data, concepts, knowledge, instructions, executable behaviors and their processing by natural and artificial intelligence. From this view, software science is theoretical software engineering; while software engineering is the applied discipline of software science in order to efficiently, economically, and reliably organize and develop large-scale software systems.

 

Neuroinformatics

An important area of basic research in cognitive informatics is neuroinformatics [Wang, 2007b], which reduces cognitive informatics theories and the studies on the internal information processing mechanisms of the brain onto the neuron and physiological level.

Definition 13. Neuroinformatics (NeI) is an interdisciplinary enquiry on the biological and physiological representation of information and knowledge in the brain at the neuron level as well as their abstract mathematical models.

In the studies of NeI, memory is recognized as the foundation, platform, as well as constraints, of any natural or machine intelligence. The cognitive models of human memory [Wang & Wang, 2006], particularly the Sensory Buffer Memory (SBM), Short-Term Memory (STM), Long-Term Memory (LTM), Action-Buffer Memory (ABM), and Conscious State Memory (CSM), reveal the fundamental mechanisms of neural informatics. In the cognitive memory structure, the ABM and CSM are newly identified by Wang and his colleagues [Wang and Wang, 2006]. To rigorously explain the hierarchical and dynamic neural cluster model of memory at physiological level, a logical model of memory is created known as the Object-Attribute-Relation (OAR) model [Wang, 2007]. Based on OAR, knowledge is formally explained as acquired information in forms of abstract knowledge, intelligence, experience and skills through learning in LTM or ABM. Therefore, knowledge as the results of learning can be rigorously represented by the OAR structure, which can be formally manipulated by concept compositions between the existing OAR and a newly created sub-OAR (sOAR).

The theories of NeI [Wang, 2011] explain a number of fundamental questions in the study of natural intelligence such as the mechanisms and the 24-hour law of long-term memory establishment. The latest development in NeI has led to the determination of the magnitude of human memory and the mechanisms of internal knowledge representation, memorization and learning. Enlightening findings in NeI are such as: a) LTM establishment is a subconscious process; b) LTM is established during sleeping; c) The major mechanism for LTM establishment is by sleeping; d) The general acquisition cycle of LTM is equal to or longer than 24 hours; e) The mechanism of LTM establishment is to update the entire memory of information represented as an OAR model in the brain; and f) Eye movement and dreams play an important role in LTM creation.

 

Brain Informatics

Definition 14. Brain informatics (BI) is a joint field of brain and information sciences that studies the information processing mechanisms of the brain by computing and medical imagination technologies.

The notion of BI is proposed in [Zhong, 2009; Wang, 2002, 2010, 2011]. A functional and logical reference model of the brain and a set of cognitive processes of the mind are systematically developed towards the exploration of the theoretical framework of BI. This work is formally represented in the Layered Reference Model of the Brain (LRMB) [Wang et al., 2006].

A fundamental challenge for almost all scientific disciplines is to explain how natural intelligence is generated by physiological organs and what the logical model of the brain is beyond its neural architectures. According to cognitive informatics and abstract intelligence, the exploration of the brain is a complicated recursive problem where contemporary denotational mathematics is needed to efficiently deal with it. Cognitive psychology and medical science are used to explain that the brain works in a certain way based on empirical observations of corresponding activities in usually overlapped brain areas. However, the lack of precise models and rigorous causality in brain studies has dissatisfied the formal expectations of researchers in computational science and mathematics, because a computer, the logical counterpart of the brain, might not be explained in such a vague and empirical approach without the support of a formal model and a rigorous means according to the abstract intelligence theories.

In order to formally explain the architectures and functions of the brain, as well as their intricate relations and interactions, systematic models of the brain are sought for revealing the principles and mechanisms of the brain at the neural, physiological, cognitive, and logical (abstract) levels. Cognitive and brain informatics investigate into the brain via not only inductive syntheses through the four cognitive levels from the bottom up in order to form theories on the basis of empirical observations, but also deductive analyses from the top down in order to explain various functional and behavioral instances according to the abstract intelligence theory.

A logical model of the brain is introduced [Wang, 2012] that maps the cognitive functions of the brain onto its neural and physiological architectures. This work leads to a coherent abstract intelligence theory based on both denotational mathematical models and cognitive psychology observations, which rigorously explains the underpinning principles and mechanisms of the brain. On the basis of the abstract intelligence theories and the logical models of the brain, a comprehensive set of cognitive behaviors as identified in LRMB such as perception, inference and learning can be rigorously explained and simulated. The logical model of the brain and the abstract intelligence theory of the natural intelligence will enable the development of cognitive computers that perceive, think and learn. The functional and theoretical difference between cognitive computers and classic computers are that the latter are data processors based on Boolean algebra and its logical counterparts; while the former are knowledge processors based on contemporary denotational mathematics. A wide range of applications of the cognitive computers have been developing in ICIC such as, inter alia, cognitive robots, cognitive learning engines, cognitive Internet, cognitive agents, cognitive search engines, cognitive translators, cognitive control systems, cognitive automobiles, cognitive medical systems and cognitive space systems.