| Author | Huang, Qiuwei |
| Call Number | AIT Thesis no.CS-00-16 |
| Subject(s) | Genetic algorithms
|
| Note | A thesis submitted in partial fulfillment of the requirements for the degree of Master of Engineering, School of Advanced Technologies |
| Publisher | Asian Institute of Technology |
| Series Statement | Thesis ; no. CS-00-16 |
| Abstract | Feature selection refers to the task of identifying and selecting a useful subset of features to be used to represent objects from a large set of often mutually redundant, possible irrelevant features with different associated measurement costs and/or risks. Selecting an "optimal" subset of variables from a set of variables is a necessary and important step in many applications, because it can allow classification algorithms to improve their prediction accuracy, shorten the learning period, and form a simple concept. Genetic Algorithm, a general and efficient search method, is used as a tool for feature selection. Two kinds of classification procedure are used to serve as the evaluation function, namely Linear Regression classification procedure and Linear Discrimination classification procedure. Coefficient of Determination based fitness function and Partial Correlation Coefficient based fitness function are the objective functions for measurement of feature subsets. Voting algorithm can reduce the error rate of classification procedure by combining the results of multiple classifiers. Two kinds of voting algorithm are proposed in this study. The main difference between them is the way by which the classifiers are generated. The first produces classifiers by using the best individuals in each generation during GA search process; and the second does it by changing the distribution of training set. Majority voting is used for generating the final results. |
| Year | 2000 |
| Corresponding Series Added Entry | Asian Institute of Technology. Thesis ; no. CS-00-16 |
| Type | Thesis |
| School | School of Advanced Technologies (SAT) |
| Department | Department of Information and Communications Technologies (DICT) |
| Academic Program/FoS | Computer Science (CS) |
| Chairperson(s) | Qi, Yulu; |
| Examination Committee(s) | Phan Minh Dung;Aekavute Sujarae; |
| Scholarship Donor(s) | Government of P.R. China; |
| Degree | Thesis (M.Eng.) - Asian Institute of Technology, 2000 |