Please use this identifier to cite or link to this item: http://scholarbank.nus.edu.sg/handle/10635/14379
Title: Soft computing techniques: Theory and application for pattern classification
Authors: GANESAN SIVAKUMAR
Keywords: fuzzy systems, genetic algorithm, rough sets, pattern classification, boosting, iterative rule learning, pittsburgh and michigan approach
Issue Date: 30-Nov-2004
Source: GANESAN SIVAKUMAR (2004-11-30). Soft computing techniques: Theory and application for pattern classification. ScholarBank@NUS Repository.
Abstract: There are a variety of real world problems such as Pattern Recognition, Image processing, Voice recognition, Data mining etc. for which, normal computing techniques are either inadequate or very tedious to apply. Soft computing techniques have been developed to fill this gap and have gained increasing popularity in the recent years. Leading examples of popular soft computing techniques are fuzzy systems, rough sets, neural networks, genetic algorithms, simulated annealing etc. In addition to solving such real world problems, soft computing techniques are also gaining acceptance in areas such as Control Systems, IP routing systems etc. where the regular computing techniques were considered to be the de-facto standard.In this thesis, basic ideas regarding Fuzzy systems, Rough Sets and Genetic algorithms are introduced, followed by the research work available in the literature. Then, the author's contribution on Pattern Classification based upon Rough set techniques, fuzzy systems and genetic algorithms are described. First, a classifier based on a combination of rough sets and NNR technique is proposed, which performs better than NNR technique alone. This is followed by a fuzzy classifier based on Pittsburgh approach genetic algorithm, in combination with the grade of certainty (CF). The classifier performs better than a Pittsburgh approach fuzzy classifier without Grade of certainty. It is also compared with a Michigan approach fuzzy classifier which neither tunes the membership functions nor minimizes the number of fuzzy rules. Finally, a new classifier based on fuzzy lower and upper approximations and a boosting technique is proposed, which makes use of the Plausibility factor (PF). Further, the Plausibility factor (PF) is compared with the grade of certainty to prove its efficacy. The performance of the classifiers are illustrated on well known test problems. In the final part, possible future directions for further research is discussed.
URI: http://scholarbank.nus.edu.sg/handle/10635/14379
Appears in Collections:Master's Theses (Open)

Show full item record
Files in This Item:
File Description SizeFormatAccess SettingsVersion 
thesis.pdf500.97 kBAdobe PDF

OPEN

NoneView/Download

Page view(s)

231
checked on Dec 11, 2017

Download(s)

468
checked on Dec 11, 2017

Google ScholarTM

Check


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.