A characterization of simple recurrent neural networks with two hidden units as a language recognizer

Azusa Iwata, Yoshihisa Shinozawa, Akito Sakurai

研究成果: Conference contribution

抄録

We give a necessary condition that a simple recurrent neural network with two sigmoidal hidden units to implement a recognizer of the formal language {a n b n | n > 0 } which is generated by a set of generating rules {S→aSb, S→ab } and show that by setting parameters so as to conform to the condition we get a recognizer of the language. The condition implies instability of learning process reported in previous studies. The condition also implies, contrary to its success in implementing the recognizer, difficulty of getting a recognizer of more complicated languages.

本文言語English
ホスト出版物のタイトルNeural Information Processing - 14th International Conference, ICONIP 2007, Revised Selected Papers
ページ436-445
ページ数10
PART 1
DOI
出版ステータスPublished - 2008
イベント14th International Conference on Neural Information Processing, ICONIP 2007 - Kitakyushu, Japan
継続期間: 2007 11 132007 11 16

出版物シリーズ

名前Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
番号PART 1
4984 LNCS
ISSN(印刷版)0302-9743
ISSN(電子版)1611-3349

Other

Other14th International Conference on Neural Information Processing, ICONIP 2007
国/地域Japan
CityKitakyushu
Period07/11/1307/11/16

ASJC Scopus subject areas

  • 理論的コンピュータサイエンス
  • コンピュータ サイエンス(全般)

フィンガープリント

「A characterization of simple recurrent neural networks with two hidden units as a language recognizer」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