| Presentation | 2005/3/23 Learning of non-i.i.d samples and model-selection Koichiro YAMAUCHI, Jiro HAYAMI, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | Incremental learning methods usually face a problem of forgetting. To avoid the problem, the system usually need to re-learn old instances again. The learning sometimes wastes long learning time. In contrast, k-Nearest Neighbors(k-NN) memorize new instances only by appending the new instances into its database so that k-NN do not waste learning time. However, k-NN waste a large amount of resources to record all instances. To solve the problem, this paper presents a model-based incremental learning system not only for function approximation but also for clustering. This method reduces appearant learning time by introducing sleep phase. Therefore, during wake phase, the system realizes recognition of known instances and memorizing unknown new instances simultaneously. On the other hand, during sleep phase, the system realizes model-selection for reduction of redundant hidden units. In this report, we propose the extended version of the above. Therefore. the new system employees the generalized regression neural network(GRNN) to memorize new instances to achieve stable learning of non independent and identical distributed (non-i.i.d) patterns. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Incremental Learning / GRBF / GRNN / Sleep / Model Selection |
| Paper # | NC2004-216 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 2005/3/23(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) | Learning of non-i.i.d samples and model-selection |
| Sub Title (in English) | |
| Keyword(1) | Incremental Learning |
| Keyword(2) | GRBF |
| Keyword(3) | GRNN |
| Keyword(4) | Sleep |
| Keyword(5) | Model Selection |
| 1st Author's Name | Koichiro YAMAUCHI |
| 1st Author's Affiliation | Graduate School of Information Science and Technology, Hokkaido University() |
| 2nd Author's Name | Jiro HAYAMI |
| 2nd Author's Affiliation | Graduate School of Information Science and Technology, Hokkaido University |
| Date | 2005/3/23 |
| Paper # | NC2004-216 |
| Volume (vol) | vol.104 |
| Number (no) | 760 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |