Keynote Lecture

Affinities Between Perceptual Granules -
Near Set Foundations and Perspectives

Prof. James F. Peters

Prof. James F. Peters

University of Manitoba, Canada

Due to unexpected medical problems of Prof. James F. Peters his lecture has been cancelled.

Abstract

This paper gives a concise overview of the foundations of a perceptual near set approach to the discovery of affinities between perceptual objects and perceptual granules that provide a basis for perceptual information systems useful in science and engineering. Granulation can be viewed as a human way of achieving data compression and it plays a key role in implementing the divide-and-conquer strategy in human problem-solving. A perceptual object is something perceptible to the senses or knowable by the mind. Perceptual objects that have the same appearance are considered to be perceptually near each other [4, 6], i.e., perceived objects are those objects that have perceived affinities or, at least, similar descriptions. Examples of perceptual objects include observable organism behaviour, growth rates, soil erosion, events containing the outcomes of experiments such as energizing a network, testing digital camera functions, microscope images, MRI scans, and the results of searches for relevant web pages. In keeping with the approach to pattern recognition suggested by Monique Pavel in 1993 [1], the features of a perceptual object are represented by probe functions or, simply, by probes. Examples of probes are functions that measure such things as conditional probability, contour, colour, shape, shape, texture, and bilateral symmetry. In general, a probe function can be thought of as a sensor.

A perceptual granule is a finite, non-empty set OF (for simplicity, O is an abbreviation for OF) containing sample perceptual objects with common descriptions constructed using a set F containing probe functions representing perceptual object features [4]. In effect, a perceptual granule is a set of perceptual objects originating from observations of the objects in the physical world. Near set theory [5, 7] provides a basis for observation, comparison and classification of perceptual granules. A perceptual information system is a real valued, total, deterministic information system, where O is a set of perceptual objects, while F a set of probe functions.

Let be a perceptual system and let X, Y c O,X != Y. Let F denote a set of functions representing features of objects in O. Set X is perceptually near Y if, and only if the perceptual objects in X and Y have matching descriptions defined by the probe functions in F. It has been shown that near sets [5] are a generalization of rough sets introduced by Zdzisław Pawlak during the early 1980s [2] and elaborated in [3]. That is, every rough set is a near set but not every near set is a rough set. It has recently been shown that the family of near sets is a chopped lattice [7]. In addition, it can be shown that pairs of fuzzy sets with non-empty cores are near sets. The connections between these three forms of sets will be briefly shown in this paper. Examples of different types of peceptual granules and perceptual systems will also be given in this chapter.

  • M. Pavel, Fundamentals of Pattern Recognition, 2nd ed., Marcel Dekker, Inc., N.Y., U.S.A., 1993.
  • Z. Pawlak, Classification of objects by means of attributes, Polish Academy of Sciences 429.
  • Z. Pawlak, A. Skowron, Rudiments of rough sets, Information Sciences 177 (2007) 3–27.
  • J. Peters, Classification of perceptual objects by means of features, Int. J. of Info. Technology & Intelligent Computing 2008 , in press.
  • J. Peters, Near sets. general theory about nearness of objects, Applied Mathematical Sciences 1 (53) (2007) 2609–2029.
  • J. Peters, Discovery of perceputally near information granules, Novel Developments in Granular Computing: Applications of Advanced Human Reasoning and Soft Computation, Information Science Reference, Hersey, N.Y., U.S.A., 2008, submitted.
  • J. Peters, P. Wasilewski, Foundations of near sets, Information Sciences (2008) Submitted.

Brief biography of the Speaker

James F. Peters, B.A., Philos. (1961), B.Sc., Math. (1965), M.Sc., Math. (1967), Ph.D., Constructive Specification of Communicating Systems (1991), Postdoctoral Fellow, Syracuse University, and Rome Laboratories (1991), Asst. Prof., University of Arkansas and Researcher in the Mission Sequencing and Telecommunications Divisions at the Jet Propulsion Laboratory/Caltech, Pasadena, California (1992-1994), is now a Full Professor in the Department of Electrical and Computer Engineering (ECE) at the University of Manitoba.

In 2002, he coauthored with Prof. Zdzisław Pawlak a poem entitled Near To [1]. This poem called attention to what it means for objects such as snowflakes, winter sky and frozen winter ground to be near each other, aesthetically. This later led to the discovery of near sets, a generalization of rough sets and an ideal framework for perceptual systems and pattern recognition. It has recently been proved that the family of near sets is a chopped lattice [2]. Since 2006, more than a dozen journal articles either directly or indirectly about nearness of objects, nearness approximation spaces, near images, nearness relations and various types of near sets have been published in refereed journals and international conferences.

In April, 2008, Dr. Peters received the International Journal of Intelligent Computing and Cybernetics best journal article award. In 2007, He received a Best Paper Award from Springer, Berlin and Joint Rough Set Symposium 2007 (JRS 2007) Program Committee, for a paper on robotic target tracking with approximation space-based feedback and approximate adaptive learning capability. In 2007, Dr. Peters was a plenary speaker on image pattern recognition and biologically-inspired adaptive learning at two international conferences (JRS 2007, Toronto and RSEISP 2007, Warsaw). He is the recipient of the IEEE Gold Medallion Award Medal (2000) and an IFAC Best Paper Award (1998) for a paper on rough control.

Currently, he research group leader in the Computational Intelligence Laboratory, University of Manitoba [3]. He also is Co-Editor-in-Chief of the Transactions on Rough Sets Journal published by Springer-Verlag [4], Co-Founder and Research Group Leader, Computational Intelligence Laboratory in the ECE Department (1996- ), current member of the Steering Committee, International Rough Sets Society [5], and current member of the Executive Board of IFAC Canada. He was also a Plenary Speaker at the Symposium on Methods of Artificial Intelligence (AIMETH 2005) where Prof. Lotfi A. Zadeh was the Honorary Chair, Keynote speaker at the International Workshop on Monitoring, Security and Rescue Techniques in Multiagent Systems (MSRAS 2004), Płock, Poland, June 2004, Plenary speaker at the 9th Int. Conf. on Rough Sets, Fuzzy Sets, Data Mining and Granular Computing (RSFDGrC 2003), October 2003, Chongqing, China, Program Co-Chair, Rough Sets and Knowledge Technology (RSKT 2006) and RSFDGrC 2005, Program Chair for North America for the 3rd Int. Conf. on Rough Sets and Current Trends in Computing (RSCTC’02), Workshop Co-Chair of COMPSAC’02 Workshop on the Foundation of Data Mining via Granular and Rough Computing, Program Committee Member of RSCTC 2000 and numerous other conferences. He served as Guest Editor for the International Journal on Intelligent Systems, 1999, 2001 and 2002. Since 1997, he has published over 300 articles in refereed journals, edited volumes, international conferences and workshops [6].

His current research interests are in near sets, perceptual systems (e.g., perceptual information systems that result from sample percepts), pattern recognition, and therapeutic telegaming in pervasive computing environments. He has also done considerable research in ethology and image processing, rough set theory, adaptive learning, biologically-inspired designs of intelligent systems (vision systems that learn), the extension of ethology (study of behaviour of biological organisms) to the study of intelligent system behaviour.

Copyright © ICMMI 2017