| Presentation | 2007-03-14 Unbiased Learning for Hierarchical Models Masashi SEKINO, Katsumi NITTA, |
|---|---|
| PDF Download Page |  PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | It is known that overfitting occurs when a conventional statistical learning method such as maximum likelihood estimation, maximum a posteriori estimation or Bayesian estimation is applied to hierarchical models. In this paper, we clarify the cause of why overfitting occurs when a conventional statistical learning method is applied to hierarchical models, and propose "Unbiased Learning" which is a learning framework based on unbiased likelihood (information criterion) for hierarchical models. We confirm the effectiveness of unbiased learning when applied to kernel regression model. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | hierarchical model / statistical learning / information criterion / model selection / over fitting |
| Paper # | NC2006-136 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2007/3/7(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) | Unbiased Learning for Hierarchical Models |
| Sub Title (in English) | |
| Keyword(1) | hierarchical model |
| Keyword(2) | statistical learning |
| Keyword(3) | information criterion |
| Keyword(4) | model selection |
| Keyword(5) | over fitting |
| 1st Author's Name | Masashi SEKINO |
| 1st Author's Affiliation | Tokyo Institute of Technology() |
| 2nd Author's Name | Katsumi NITTA |
| 2nd Author's Affiliation | Tokyo Institute of Technology |
| Date | 2007-03-14 |
| Paper # | NC2006-136 |
| Volume (vol) | vol.106 |
| Number (no) | 588 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |