Fuzzy logic of Aristotelian forms
Abstract
Modelbased approaches to pattern recognition and machine vision have been proposed to overcome the exorbitant training requirements of earlier computational paradigms. However, uncertainties in data were found to lead to a combinatorial explosion of the computational complexity. This issue is related here to the roles of a priori knowledge vs. adaptive learning. What is the apriori knowledge representation that supports learning? I introduce Modeling Field Theory (MFT), a modelbased neural network whose adaptive learning is based on a priori models. These models combine deterministic, fuzzy, and statistical aspects to account for a priori knowledge, its fuzzy nature, and data uncertainties. In the process of learning, a priori fuzzy concepts converge to crisp or probabilistic concepts. The MFT is a convergent dynamical system of only linear computational complexity. Fuzzy logic turns out to be essential for reducing the combinatorial complexity to linear one. I will discuss the relationship of the new computational paradigm to two theories due to Aristotle: theory of Forms and logic. While theory of Forms argued that the mind cannot be based on readymade a priori concepts, Aristotelian logic operated with just such concepts. I discuss an interpretation of MFT suggesting that its fuzzy logic, combining apriority andmore »
 Authors:

 Nichols Research Corp., Lexington, MA (United States)
 Publication Date:
 OSTI Identifier:
 466430
 Report Number(s):
 CONF9610138
TRN: 97:0013090009
 Resource Type:
 Conference
 Resource Relation:
 Conference: International multidisciplinary conference on intelligent systems: a semiotic perspective, Gaithersburg, MD (United States), 2123 Oct 1996; Other Information: PBD: 1996; Related Information: Is Part Of Intelligent systems: A semiotic perspective. Volume I: Theoretical semiotics; Albus, J.; Meystel, A.; Quintero, R.; PB: 303 p.
 Country of Publication:
 United States
 Language:
 English
 Subject:
 99 MATHEMATICS, COMPUTERS, INFORMATION SCIENCE, MANAGEMENT, LAW, MISCELLANEOUS; FUZZY LOGIC; ARTIFICIAL INTELLIGENCE; MATHEMATICAL LOGIC; LEARNING; NEURAL NETWORKS; PATTERN RECOGNITION
Citation Formats
Perlovsky, L I. Fuzzy logic of Aristotelian forms. United States: N. p., 1996.
Web.
Perlovsky, L I. Fuzzy logic of Aristotelian forms. United States.
Perlovsky, L I. Tue .
"Fuzzy logic of Aristotelian forms". United States.
@article{osti_466430,
title = {Fuzzy logic of Aristotelian forms},
author = {Perlovsky, L I},
abstractNote = {Modelbased approaches to pattern recognition and machine vision have been proposed to overcome the exorbitant training requirements of earlier computational paradigms. However, uncertainties in data were found to lead to a combinatorial explosion of the computational complexity. This issue is related here to the roles of a priori knowledge vs. adaptive learning. What is the apriori knowledge representation that supports learning? I introduce Modeling Field Theory (MFT), a modelbased neural network whose adaptive learning is based on a priori models. These models combine deterministic, fuzzy, and statistical aspects to account for a priori knowledge, its fuzzy nature, and data uncertainties. In the process of learning, a priori fuzzy concepts converge to crisp or probabilistic concepts. The MFT is a convergent dynamical system of only linear computational complexity. Fuzzy logic turns out to be essential for reducing the combinatorial complexity to linear one. I will discuss the relationship of the new computational paradigm to two theories due to Aristotle: theory of Forms and logic. While theory of Forms argued that the mind cannot be based on readymade a priori concepts, Aristotelian logic operated with just such concepts. I discuss an interpretation of MFT suggesting that its fuzzy logic, combining apriority and adaptivity, implements Aristotelian theory of Forms (theory of mind). Thus, 2300 years after Aristotle, a logic is developed suitable for his theory of mind.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = {1996},
month = {12}
}