| Presentation | 1994/5/19 Increase the Capability of MLP Using Cross-Layer Connections Qiangfu Zhao, |
|---|---|
| PDF Download Page | PDF download Page Link |
| Abstract(in Japanese) | (See Japanese page) |
| Abstract(in English) | In the study of multilayer perceptrons(MLP),a common assumption by all researchers is that no cross-layer connections exist in the network.Although learning of MLP under this assumption becomes very simple,the capability of MLP is also limited.For highly nonlinear decision problems,if the number of neurons in a certain layer is not large enough,samples of different patterns may be mapped into one point in the next layer,and this error can not be corrected in the following layers regardless how many layers we use.This problem however,can be solved simply by using cross-layer connections.In the MLP with cross-layer connections(CLC-MLP),part of the data used in a lower layer can also be used in the higher layers,and classification errors made in a layer can be corrected in the higher layers by using these data appropriately.In this paper,the structure,the neuron models and the learning rules for CLC-MLP are studied,and several new ideas are proposed. |
| Keyword(in Japanese) | (See Japanese page) |
| Keyword(in English) | Multi-Layer Perceptron / Pattern Recognition / Competitive Learing / Self-Organization / Cross-Layer Connections |
| Paper # | NC94-8 |
| Date of Issue |
| Conference Information | |
| Committee | NC |
|---|---|
| Conference Date | 1994/5/19(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 | ENG |
| Title (in Japanese) | (See Japanese page) |
| Sub Title (in Japanese) | (See Japanese page) |
| Title (in English) | Increase the Capability of MLP Using Cross-Layer Connections |
| Sub Title (in English) | |
| Keyword(1) | Multi-Layer Perceptron |
| Keyword(2) | Pattern Recognition |
| Keyword(3) | Competitive Learing |
| Keyword(4) | Self-Organization |
| Keyword(5) | Cross-Layer Connections |
| 1st Author's Name | Qiangfu Zhao |
| 1st Author's Affiliation | Graduate School of Information Sciences,Tohoku University() |
| Date | 1994/5/19 |
| Paper # | NC94-8 |
| Volume (vol) | vol.94 |
| Number (no) | 40 |
| Page | pp.pp.- |
| #Pages | 6 |
| Date of Issue | |