頁籤選單縮合
題名 | Fuzzy Rule Extraction by a Hybrid GA and SVD-QR Method for Pattern Classification= |
---|---|
作者 | Wong, Ching-chang; Lin, Bo-chen; Chen, Chia-chong; |
期刊 | International Journal of Electrical Engineering |
出版日期 | 20050500 |
卷期 | 12:2 民94.05 |
頁次 | 頁113-123 |
分類號 | 448.942 |
語文 | eng |
關鍵詞 | Fuzzy sytem; Pattern Classification; Genetic algorithms; SVD-QR method; |
英文摘要 | A hybrid Genetic Algorithm (GA) SVD-QR method is proposed for pattern classification to select an appropriate fuzzy system to minimize the number of incorrectly classified patterns and minimize the number of fuzzy rules. First, an individual in a population of GA is considered to automatically generate a rough fuzzy system. Then, an orthogonal transformation technique, known as the singular value decomposition and QR with column pivoting (SVD-QR) method, is considered to select the dominant fuzzy rules from this rough rule base. In the SVD-QR method, a firing strength matrix determined by the membership function values of the premise fuzzy sets is constructed. Then, the number of significant fuzzy rules is determined by the singular value decomposition (SVD) according to the firing strength matrix, and the significant fuzzy rules are selected by the SVD-QR method. In GA, a fitness function is proposed to guide the search procedure to select an appropriate fuzzy system such that the number of incorrectly classified patterns and the number of fuzzy rules are simultaneously minimized. Finally, two classification problems are considered to illustrate the effectiveness of the proposed approach. |
本系統之摘要資訊系依該期刊論文摘要之資訊為主。