| Presentation | 2006-01-24 Embedding of Labeled Multimodal Data Yuki SHINADA, Masashi SUGIYAMA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In order to improve the recognition accuracy of high dimensional patterns, it is important to appropriately reduce the number of dimensions of the data. We discuss the linear dimensionality reduction problem in supervised learning. The Fisher Criterion is a standard criterion for linear dimensionality reduction. It reduces dimensionality while keeping the mean distance between classes large and the variance within the class small. However, when the data of each class is multimodal, the Fisher Criterion is not able to reduce dimensionality appropriately since the mean distance between classes and the within-class variance are not well evaluated. Recently, Locality Preserving Projection (LPP) has been proposed, which reduces dimensionality while preserving local structure of the data. LPP is able to perform dimensionality reduction with the multimodality of the data preserved. However, since LPP is an unsupervised method, it is not necessarily effective for pattern classification. In this paper, we therefore propose a new supervised linear dimensionality reduction method which preserves local structure and takes the class information into account. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | dimensionality reduction / feature extraction / labeled multimodal data / Fisher Criterion / Locality Preserving Criterion / Locality and Separability Preserving Criterion |
| Paper # | NC2005-102 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2006/1/17(1days) |
| Place (in Japanese) | (See Japanese page) |
| Place (in English) | |
| Topics (in Japanese) | (See Japanese page) |
| Topics (in English) | |
| Chair | |
| Vice Chair | |
| Secretary | |
| Assistant | |
| Paper Information | |
| Registration To | Neurocomputing (NC) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Embedding of Labeled Multimodal Data |
| Sub Title (in English) | |
| Keyword(1) | dimensionality reduction |
| Keyword(2) | feature extraction |
| Keyword(3) | labeled multimodal data |
| Keyword(4) | Fisher Criterion |
| Keyword(5) | Locality Preserving Criterion |
| Keyword(6) | Locality and Separability Preserving Criterion |
| 1st Author's Name | Yuki SHINADA |
| 1st Author's Affiliation | Department of Computer Science, Tokyo Institute of Technology() |
| 2nd Author's Name | Masashi SUGIYAMA |
| 2nd Author's Affiliation | Department of Computer Science, Tokyo Institute of Technology |
| Date | 2006-01-24 |
| Paper # | NC2005-102 |
| Volume (vol) | vol.105 |
| Number (no) | 544 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |