| Presentation | 2013-09-02 Optimization for Robust Support Vector Machine Utilizing Homotopy Approach Shinya SUZUMURA, Kohei OGAWA, Ichiro TAKEUCHI, Masashi SUGIYAMA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In this paper, we propose a novel optimization method for robust SVM that has robustness to outliers. In robust SVM, it is important to adaptively determine which instances are considered as outliers, and how much the effect of outliers should be alleviated. We introduce a pair of tuning parameters that adjust the outlier's criterion and effects, and develop an algorithm that computes the path of solutions when those parameters are continuously changed. Since the loss function of robust SVM is non-convex, our proposed algorithm computes a path of local optimal solutions. Unlike solution paths of convex problems, our local optimal solution path are shown to have finite number of discrete points. Our algorithm can identify those discrete points and properly handle those discontinuities in the path computation. We illustrate the effectiveness of our algorithm through numerical experiments. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Robust Support Vector Machine / Homotopy Approach / Annealing |
| Paper # | PRMU2013-36,IBISML2013-16 |
| Date of Issue |
| Conference Information | |
| Committee | PRMU |
|---|---|
| Conference Date | 2013/8/26(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) | Optimization for Robust Support Vector Machine Utilizing Homotopy Approach |
| Sub Title (in English) | |
| Keyword(1) | Robust Support Vector Machine |
| Keyword(2) | Homotopy Approach |
| Keyword(3) | Annealing |
| 1st Author's Name | Shinya SUZUMURA |
| 1st Author's Affiliation | Nagoya Institute of Technology() |
| 2nd Author's Name | Kohei OGAWA |
| 2nd Author's Affiliation | Nagoya Institute of Technology |
| 3rd Author's Name | Ichiro TAKEUCHI |
| 3rd Author's Affiliation | Nagoya Institute of Technology |
| 4th Author's Name | Masashi SUGIYAMA |
| 4th Author's Affiliation | Tokyo Institute of Technology |
| Date | 2013-09-02 |
| Paper # | PRMU2013-36,IBISML2013-16 |
| Volume (vol) | vol.113 |
| Number (no) | 196 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |