| Presentation | 2010-09-05 Behavior of kernel mutual subspace method with respect to parameters Seiji HOTTA, Tomokazu KAWAHARA, Osamu YAMAGUCHI, Hitoshi SAKANO, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Optimizing parameters of kernel methods is an unsolved problem. We report the experimental evaluation and the consideration of parameters dependencies on kernel mutual subspace method (KMS). The following parameters of KMS are considered: Gaussian kernel parameters, dimensionalities of dictionary and input subspaces, and the number of canonical angles. We evaluate recognition accuracies of KMS through experiments on ETH-80 animals database. According to searching optimal parameters exhaustively, we obtain 100% recognition accuracy, and some experimental results suggests the relationships between the dimensionality of subspaces and the degrees of freedom for object motions. Such results imply that KMS achieves high recognition rate for object recognition with optimized parameters. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | kernel mutual subspace method / object recognition / subspace method / kernel method / statistical pattern recognition / machine leaning |
| Paper # | PRMU2010-57,IBISML2010-29 |
| Date of Issue |
| Conference Information | |
| Committee | PRMU |
|---|---|
| Conference Date | 2010/8/29(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 | Pattern Recognition and Media Understanding (PRMU) |
|---|---|
| Language | JPN |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Behavior of kernel mutual subspace method with respect to parameters |
| Sub Title (in English) | |
| Keyword(1) | kernel mutual subspace method |
| Keyword(2) | object recognition |
| Keyword(3) | subspace method |
| Keyword(4) | kernel method |
| Keyword(5) | statistical pattern recognition |
| Keyword(6) | machine leaning |
| 1st Author's Name | Seiji HOTTA |
| 1st Author's Affiliation | The Graduate School of Engineering, Tokyo University of Agriculture and Technology() |
| 2nd Author's Name | Tomokazu KAWAHARA |
| 2nd Author's Affiliation | Corporate R&D Center, Toshiba Corporation |
| 3rd Author's Name | Osamu YAMAGUCHI |
| 3rd Author's Affiliation | Power and Industrial System R&D Center, Toshiba Corporation Power Systems Company |
| 4th Author's Name | Hitoshi SAKANO |
| 4th Author's Affiliation | NTT Communication Science Lab. |
| Date | 2010-09-05 |
| Paper # | PRMU2010-57,IBISML2010-29 |
| Volume (vol) | vol.110 |
| Number (no) | 187 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |