Presentation | 1994/5/19 A quasi-competitive network with unlearning Toshiki Kindo, |
---|---|
PDF Download Page | PDF download Page Link |
Abstract(in Japanese) | (See Japanese page) |
Abstract(in English) | A quasi-competitive network(QCNet)is a neural network model to approximate nonlinear functions.As the total activation of a hidden layer is fixed,QCNet gives a high performance on the interpolation between the given data.The learning is very fast because QCNet adapts its model size to a target function by creating new units.In this paper the author proposes the generalized learning algorithm which includes unlearning.QCNet is a model which represents the function from local infomation.The unlearning algorithm is simple because it needs only local informations of the model.This generalized learning algorithm suppresses the number of active units,but dosen′t effect the outpu t error and the learning speed. |
Keyword(in Japanese) | (See Japanese page) |
Keyword(in English) | quasi-competitive / unlearning / local distributed upresentation |
Paper # | NC94-2 |
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 | JPN |
Title (in Japanese) | (See Japanese page) |
Sub Title (in Japanese) | (See Japanese page) |
Title (in English) | A quasi-competitive network with unlearning |
Sub Title (in English) | |
Keyword(1) | quasi-competitive |
Keyword(2) | unlearning |
Keyword(3) | local distributed upresentation |
1st Author's Name | Toshiki Kindo |
1st Author's Affiliation | Matsushita Research Institute,Tokyo() |
Date | 1994/5/19 |
Paper # | NC94-2 |
Volume (vol) | vol.94 |
Number (no) | 40 |
Page | pp.pp.- |
#Pages | 8 |
Date of Issue |