融入叢集分析之特徵萃取法在高維度資料辨識的應用
| Autor: | Cheng-Wei Chan, 詹正維 |
|---|---|
| Rok vydání: | 2006 |
| Druh dokumentu: | 學位論文 ; thesis |
| Popis: | 94 When classify high dimensional data, the sample size is too small cause the singularity and decreasing accuracy. For processing high dimensional data, feature extraction and selection are often using to solve the small sample size problems. There are two problems for high dimensional data. First, there do often exist some differences in samples with the same label. Second, samples with different label are often overlapped. The above two problems are account for the classifier accuracy decreasing in classifying. In this research, a cluster analysis method fuse to feature extraction is proposed. The samples with the same label divide into several clusters. Because the criterion of linear discriminant analysis and nonparametric weighted feature extraction is using the ratio of between and within scatter matrix. So, the proposed method can solve the above problems. |
| Databáze: | Networked Digital Library of Theses & Dissertations |
| Externí odkaz: |
|