|Author:||Ho, Kwok-sze Keith|
|Title:||Learning local feature weights for case-base reduction|
Hong Kong Polytechnic University -- Dissertations
Department of Computing
|Pages:||vii, 127 leaves : ill. (some col.) ; 30 cm|
|Abstract:||Case-Based Reasoning (CBR) is a recent approach to problem solving and learning that has aroused much attention over the last few years. The basic idea behind this approach is to use previous cases (past experiences) to solve new and different problems. Now, the field is of widespread interest as it is more intuitive to most people including experts in solving problems. Since the field is still in the stage of highly active research, we have yet found so many unsolved problems. Among those problems, Case-Based Maintenance is worth more attention. In fact, as more and more large-scale commercial CBR systems emerge, performance of CBR systems becomes an important issue. Generally speaking, the number of cases stored in the case library of a case-based expert system is directly related to the retrieval efficiency. Although more cases in the library can improve the coverage of the problem space, the system performance will be downgraded if the size of the library grows to an unacceptable level. In this research project, we have devised a new way to reduce the size of huge case libraries so as to improve the efficiency and at the same time maintain the accuracy of the CBR System. To achieve this, we have adopted the local feature weights approach, proposed by Dr. Simon Shiu, to find out the best representatives and remove the redundant cases under an acceptable error-level. Conceptually, this approach consists of three phases. The first phase is to partition the case-base into different clusters according to the solutions of cases. The second phase is to learn the optimal local feature weights for each case in the case-base and the final phase is to reduce the case-base based on the optimal local weights. Our studies are mainly based on the last two phases. To justify the usefulness of the method, we have performed two well-designed experiments by using two different data-sets. In these two experiments, we used efficiency, competence, and ability to solve new problems as the benchmark to verify our design. To our satisfaction, the training accuracy and testing accuracy of the reduced case-bases attained 100% and 85% respectively with obvious reduction rates. In this report, this appealing method will be introduced and the results will be presented step by step.|
|Rights:||All rights reserved|
Files in This Item:
|b15995914.pdf||For All Users (off-campus access for PolyU Staff & Students only)||6.28 MB||Adobe PDF||View/Open|
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- 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.
- 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.
Please use this identifier to cite or link to this item: