Multiplication units in feedforward neural networks and its training

Dazi Li, K. Hirasawa, Jinglu Hu, J. Murata

研究成果: Chapter in Book/Report/Conference proceedingConference contribution

6 被引用数 (Scopus)

抄録

This paper proposes the application of neural networks with multiplication units to parity-N problem, mirror symmetry problem and a function approximation problem. It is clear that, higher-order terms in neural networks, such as sigma-pi unit, can improve the computational power of neural networks considerably. But how the real neurons do this is still unclear. We have used one multiplication unit to construct full higher-order terms of all the inputs, which was proved very efficient for parity-N problem. Our earlier work on applying multiplication units to other problems suffered from the drawback of gradient-based algorithm, such as backpropagation algorithms, for being easy to stuck at local minima due to the complexity of the network. In order to overcome this problem we consider a novel random search, RasID, for the training of neural networks with multiplication units, which does an intensified search where it is easy to find good solutions locally and a diversified search to escape from local minima under a pure random search scheme. The method shows its advantage on the training of neural networks with multiplication units.

本文言語英語
ホスト出版物のタイトルICONIP 2002 - Proceedings of the 9th International Conference on Neural Information Processing
ホスト出版物のサブタイトルComputational Intelligence for the E-Age
編集者Jagath C. Rajapakse, Xin Yao, Lipo Wang, Kunihiko Fukushima, Soo-Young Lee
出版社Institute of Electrical and Electronics Engineers Inc.
ページ75-79
ページ数5
ISBN(電子版)9810475241, 9789810475246
DOI
出版ステータス出版済み - 1 1 2002
イベント9th International Conference on Neural Information Processing, ICONIP 2002 - Singapore, シンガポール
継続期間: 11 18 200211 22 2002

出版物シリーズ

名前ICONIP 2002 - Proceedings of the 9th International Conference on Neural Information Processing: Computational Intelligence for the E-Age
1

その他

その他9th International Conference on Neural Information Processing, ICONIP 2002
国/地域シンガポール
CitySingapore
Period11/18/0211/22/02

All Science Journal Classification (ASJC) codes

  • コンピュータ ネットワークおよび通信
  • 情報システム
  • 信号処理

フィンガープリント

「Multiplication units in feedforward neural networks and its training」の研究トピックを掘り下げます。これらがまとまってユニークなフィンガープリントを構成します。

引用スタイル