Cognitive Sciences Applications in Big Data (General Joint Session at WMSCI 2014)
- Duration: 41:34
- Updated: 17 Aug 2014
"Cognitive Sciences Applications in Big Data"
(General Joint Session at WMSCI 2014)
Dr. Leonid Perlovsky
Harvard University and The Air Force Research Laboratory, USA
Abstract:
Big Data problems have been efficiently addressed with cognitive algorithms modeling mechanisms of the mind. The talk describes cognitive algorithms, their applications to various engineering problems, including Big Data, and their foundations in mathematical models of the mind including higher cognitive abilities. Mechanisms of the mind include concepts, emotions, hierarchy, dynamic logic, and interaction between language and cognition. Big Data analytics requires algorithms modeling all these abilities. Machine learning, artificial intelligence, and modeling of the mind has been plagued by computational complexity since the 1960s. Dynamic logic overcomes computational complexity when analyzing Big Data. It is a process-logic, which replaces classical logic; it serves as a basis for cognitive algorithms and for a mathematical theory of learning, combining the mechanisms of the mind into a hierarchical system of mental processes. Each process proceeds "from vague to crisp," from vague representation-concepts to crisp ones. Brain imaging experiments (Bar et al 2006; Kveraga et al 2007) confirmed this as an adequate model of the brain perception and cognition.
Computational difficulty is related to Gödelian problems in logic: computational complexity is a manifestation of Gödelian incompleteness in finite systems, such as computers or brains. The mind is "not logical." Dynamic logic overcomes this difficulty. Engineering applications demonstrate orders of magnitude improvement in Big Data analytics, data mining, information integration, financial predictions, genetic studies, cybersecurity.
The talk presents the dual hierarchy model of interactions between language and cognition. It enables integrating language, text, and sensor data. A number of "mysteries" in this interaction are explained: what is the difference between them; what is the role of language in cognition, why children can talk before they really understand, how much adults are different from children in this respect, etc. These are explained in the model, and explanations are confirmed in brain imaging experiments (Binder et al 2005; Price 2012). Much difficulties in developing Big Data algorithms are related to confusing language and cognition.
The knowledge instinct drives acquisition of cognitive ability and is a foundation of all our higher cognitive abilities. Its satisfaction is experienced as aesthetic emotions (experimentally confirmed in Cabanac et al 2010). Efficient engineering algorithms must model these emotional abilities (Perlovsky, Deming, Ilin, 2011). The hierarchy of aesthetic emotions is discussed from understanding of everyday objects, to understanding of abstract concepts throughout the hierarchy, to the near top of the mental hierarchy. Contents of these "highest" concepts are discussed and the corresponding aesthetic emotions are related to the beautiful. Experimental tests of this conjecture are for the near future.
Contradictions among knowledge are experienced as negative aesthetic emotions, cognitive dissonance. Development of robots and human-computer interactions require algorithms modeling this ability. Cognitive dissonance counteracts the knowledge instinct and would prevent accumulation of knowledge and the entire human evolution, if not a special ability evolved for overcoming these emotions. It follows from the dual hierarchy model that this mechanism is music. This theoretical prediction has been experimentally confirmed (Masataka et al 2012, 2013, Cabanac et al, 2013). This explains the origin and evolution of music, what Darwin called the greatest mystery.
http://wn.com/Cognitive_Sciences_Applications_in_Big_Data_(General_Joint_Session_at_WMSCI_2014)
"Cognitive Sciences Applications in Big Data"
(General Joint Session at WMSCI 2014)
Dr. Leonid Perlovsky
Harvard University and The Air Force Research Laboratory, USA
Abstract:
Big Data problems have been efficiently addressed with cognitive algorithms modeling mechanisms of the mind. The talk describes cognitive algorithms, their applications to various engineering problems, including Big Data, and their foundations in mathematical models of the mind including higher cognitive abilities. Mechanisms of the mind include concepts, emotions, hierarchy, dynamic logic, and interaction between language and cognition. Big Data analytics requires algorithms modeling all these abilities. Machine learning, artificial intelligence, and modeling of the mind has been plagued by computational complexity since the 1960s. Dynamic logic overcomes computational complexity when analyzing Big Data. It is a process-logic, which replaces classical logic; it serves as a basis for cognitive algorithms and for a mathematical theory of learning, combining the mechanisms of the mind into a hierarchical system of mental processes. Each process proceeds "from vague to crisp," from vague representation-concepts to crisp ones. Brain imaging experiments (Bar et al 2006; Kveraga et al 2007) confirmed this as an adequate model of the brain perception and cognition.
Computational difficulty is related to Gödelian problems in logic: computational complexity is a manifestation of Gödelian incompleteness in finite systems, such as computers or brains. The mind is "not logical." Dynamic logic overcomes this difficulty. Engineering applications demonstrate orders of magnitude improvement in Big Data analytics, data mining, information integration, financial predictions, genetic studies, cybersecurity.
The talk presents the dual hierarchy model of interactions between language and cognition. It enables integrating language, text, and sensor data. A number of "mysteries" in this interaction are explained: what is the difference between them; what is the role of language in cognition, why children can talk before they really understand, how much adults are different from children in this respect, etc. These are explained in the model, and explanations are confirmed in brain imaging experiments (Binder et al 2005; Price 2012). Much difficulties in developing Big Data algorithms are related to confusing language and cognition.
The knowledge instinct drives acquisition of cognitive ability and is a foundation of all our higher cognitive abilities. Its satisfaction is experienced as aesthetic emotions (experimentally confirmed in Cabanac et al 2010). Efficient engineering algorithms must model these emotional abilities (Perlovsky, Deming, Ilin, 2011). The hierarchy of aesthetic emotions is discussed from understanding of everyday objects, to understanding of abstract concepts throughout the hierarchy, to the near top of the mental hierarchy. Contents of these "highest" concepts are discussed and the corresponding aesthetic emotions are related to the beautiful. Experimental tests of this conjecture are for the near future.
Contradictions among knowledge are experienced as negative aesthetic emotions, cognitive dissonance. Development of robots and human-computer interactions require algorithms modeling this ability. Cognitive dissonance counteracts the knowledge instinct and would prevent accumulation of knowledge and the entire human evolution, if not a special ability evolved for overcoming these emotions. It follows from the dual hierarchy model that this mechanism is music. This theoretical prediction has been experimentally confirmed (Masataka et al 2012, 2013, Cabanac et al, 2013). This explains the origin and evolution of music, what Darwin called the greatest mystery.
- published: 17 Aug 2014
- views: 133