Name
Abstract | ||
|---|---|---|
Recently, Locality Sensitive Discriminant Analysis (LSDA) has been proposed as an efficient feature extraction approach. By analyzing the local manifold structure of high-dimensional data, LSDA can obtain a subspace in which the nearby points with the same label are close to each other while the nearby points with different labels are far apart. However, because LSDA only takes the local information into consideration, it may fail to deal with the data set which contains some outliers. In order to address this limitation, a new algorithm called Improved Locality Sensitive Discriminant Analysis (ILSDA) is proposed in this paper. By integrating the intra-class scatter matrix into our algorithm, ILSDA can not only preserve the local discriminant neighborhood structure of the data, but also pull the outlier samples more close to their class centers, which makes it outperform the original LSDA and some other state of the art algorithms. Extensive experimental results on several publicly available image datasets show the feasibility and effectiveness of our proposed approach. |
Year | DOI | Venue |
|---|---|---|
2015 | 10.1007/s11042-013-1429-5 | Multimedia Tools Appl. |
Keywords | Field | DocType |
Feature extraction,Outliers,ILSDA,Face image recognition | Manifold structure,Data mining,Locality sensitive discriminant analysis,Subspace topology,Pattern recognition,Computer science,Discriminant,Outlier,Feature extraction,Artificial intelligence,Scatter matrix | Journal |
Volume | Issue | ISSN |
74 | 1 | 1380-7501 |
Citations | PageRank | References |
2 | 0.36 | 25 |
Authors | ||
4 | ||
Name | Order | Citations | PageRank |
|---|---|---|---|
| Yugen Yi | 1 | 92 | 15.25 |
| Baoxue Zhang | 2 | 142 | 11.16 |
| Jun Kong | 3 | 158 | 14.14 |
| Jianzhong Wang | 4 | 214 | 17.72 |