Learning global feature weights for clustering and classification of cases

Pao Yue-kong Library Electronic Theses Database

Learning global feature weights for clustering and classification of cases

 

Author: Lam, Bun Martin
Title: Learning global feature weights for clustering and classification of cases
Degree: M.Sc.
Year: 2001
Subject: Case-based reasoning
Hong Kong Polytechnic University -- Dissertations
Department: Multi-disciplinary Studies
Dept. of Computing
Pages: vii, 67 leaves : ill. (some col.) ; 30 cm
Language: English
InnoPac Record: http://library.polyu.edu.hk/record=b1599590
URI: http://theses.lib.polyu.edu.hk/handle/200/901
Abstract: Case-Based Reasoning (CBR) systems have increased awareness in the last few years. A CBR system is a means to store, share and re-use experience. CBR is based on the idea that new problems can often be solved by using past solutions. The basic method used to implement CBR is to build a case-base of previously solved problems. These cases are then retrieved and adapted to solve new problems. However, when the case-base is growing up to a large scale, the Case-Base Maintenance (CBM) issue will become very important. This project presents a CBM methodology for clustering and classification of cases in case-base systems from scratch, which uses little or no domain-specific knowledge, consists of two phases. The first phase is to evaluate feature importance, i.e., learning global feature weights from the case-base. This algorithm can recognize salient features and eliminate irrelevant features from the case-base. Its results are helpful to provide a reasonable clustering of the case-base. The second phase is to partition the case-base; the clusters identified convey different concepts within the case-base. Each of these concepts signifies a subset of the problem domain that differs characteristically from the rest of the domain. The effectiveness of the method is demonstrated experimentally on Boston Housing Data and Pima Indians Diabetes Database. Experimental analysis of our methodology shows promising results: the performance of clustering with learned feature weights is much better than the performance without feature weights.

Files in this item

Files Size Format
b15995902.pdf 3.590Mb PDF
Copyright Undertaking
As a bona fide Library user, I declare that:
  1. I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
  2. I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
  3. I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.

     

Quick Search

Browse

More Information