Date of Submission

3-28-1997

Date of Award

3-28-1998

Institute Name (Publisher)

Indian Statistical Institute

Document Type

Doctoral Thesis

Degree Name

Doctor of Philosophy

Subject Name

Computer Science

Department

Electronics and Communication Sciences Unit (ECSU-Kolkata)

Supervisor

Ray, Kumar Sankar (ECSU-Kolkata; ISI)

Abstract (Summary of the Work)

In real world, pattern classification and object recognition problems are faced with fuzzi- ness that is connected with diverse facets of cognitive activity of the human being. An origin of sources of fuzziness is related to labels expressed in feature space as well as to labels of classes taken into account in classification and /or recognition procedures. Though a lot of scientific efforts have already been dedicated to pattern recognition problems, especially to classification procedures, still pattern recognition is confronted with a continuous challenge coming from a human being who can perform lot of ex- tremely complex classification tasks by some sort of mental reasoning which can not be represented, in straightforward way, through computer algorithm. But, fuzzy set provides a plausible tool for modeling and mimicking cognitive processes of the human reasoning, especially those concerning recognition aspects. Fuzzy reasoning, proposed by Zadeh is one such tool. So far we have seen many successful applications of fuzzy reasoning [54,61,62,108] to the design of fuzzy logic controllers (8,70].But, fuzzy reasoning approach can also be applied very successfully to pattern classification (33,34,35,36,115] and occluded object recognition problems where multi- ple classifications of a pattern and/or object is desired. In conventional approaches topattern classification/ object recognition [55,56,57,58,85,86,87,97] such multiple classifi- cation, which is essentially needed due to the overlap of features of the patterns/ objects, is not considered.There are two main approaches to pattern classification, namely the decision the- oretic and the syntactic. Since, the fuzzy reasoning approach to pattern classification is similar to the decision theoretic method of pattern classification, we will first briefly describe the basic concept of the decision theoretic approach to pattern classification and then subsequently try to establish the similarity between the decision theoretic approach and the fuzzy reasoning approach to pattern classification.Under decision theoretic approach, each pattern is represented by a vector of features. The pattern space is divided into a number of regions, each of which represents a prototype pattern or a cluster of patterns. A decision function maps the given patterns to previously determined regions.In the fuzzy reasoning approach to pattern classification each element of the feature vector / pattern vector is represented by the fuzzy linguistic variable instead of a real number. For instance, suppose we have a (2x1) feature vector / pattern vector F %3D (F1, F1)T, T is transpose where F1 is the first formant frequency of a speech signal and F2, is the second formant frequency. In the conventional decision theoretic approach to pattern classification, F1 and F2 are two features and are represented by, say, 800 Hz and 550 Hz. Whereas in the fuzzy reasoning approach to pattern classification F and F, are represented by the fuzzy linguistic variables, e.g., F1 is small and F2 is medium. The elements of the feature vector / pattern vector constitute the antecedent part of the fuzzy implication. The consequent part of the fuzzy implication is a fuzzy set which represents the possibility of occurrence of different classes of patterns on the pattern space. Thus, fuzzy If Then rules, which map the given patterns to previously determined regions, may be used for pattern classification / object recognition problems.

Comments

ProQuest Collection ID: http://gateway.proquest.com/openurl?url_ver=Z39.88-2004&rft_val_fmt=info:ofi/fmt:kev:mtx:dissertation&res_dat=xri:pqm&rft_dat=xri:pqdiss:28842940

Control Number

ISILib-TH280

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

DOI

http://dspace.isical.ac.in:8080/jspui/handle/10263/2146

Download

Included in

Mathematics Commons

Share

COinS