ニューラルネットワークによるパズルの求解 : ホップフィールドネットワークで数独は解ける(社会システムと知能)  [in Japanese] Solving Puzzle with Neural Network : Hopfield Network can Solve Sudoku  [in Japanese]

    • 大谷 哲広 OHTANI Akihiro
    • 日本大学大学院生産工学研究科 Department of Mathematical Information Engineering, Graduate School of Industrial Technology, Nihon University
    • 松田 聖 MATSUDA Satoshi
    • 日本大学生産工学部 Department of Mathematical Information Engineering, College of Industrial Technology, Nihon University

Abstract

ホップフィールドネットワークで数独を解くことを試み,挑戦したすべての問題で正しい解を得ることができた.これまで,ホップフィールドネットワークを用いた最適化問題へ挑戦が多く試みられているが,それらの試みはいずれも十分な成果をあげられていない.本研究で挑戦した200問以上の問題のほとんどは,人間にとって大変難しいものであり,これまで正解を得られなかったことがない.人間のすべての知的活動は最適化過程であるという仮説に基づき,筆者らは意志決定のホップフィールドネットワークモデルを提案しているが,パズルを解くことも典型的な知的活動である.最後に数独の解法及びシミュレーションに基づき,人間とホップフィールドネットワークの解決能力における類似性や相違性も考える.

We show that Hopfield network can solve all the Sudoku puzzles we have attacked so far. Most of them are very difficult for us to solve. There have been many attempts at solving optimization problems and puzzles with Hopfield networks, however, they have resulted in unsatisfactory performance. So, although there is no proof, the perfect performance shown through simulations here is remarkable. On the other hand, based on the insight that all intellectual activities of human beings are performed as optimization processes, we previously proposed Hopfield network models of decision making. Solving puzzles is also a typical intellectual activity. Some considerations are also made on the similarities and/or differences between human beings and Hopfield networks in problem solving abilities.

Journal

IEICE technical report. Artificial intelligence and knowledge-based processing   [List of Volumes]

IEICE technical report. Artificial intelligence and knowledge-based processing 108(456), 141-146, 2009-02-23  [Table of Contents]

The Institute of Electronics, Information and Communication Engineers

References:  16

You must have a user ID to see the references.If you already have a user ID, please click "Login" to access the info.New users can click "Sign Up" to register for an user ID.

Cited by:  1

You must have a user ID to see the cited references.If you already have a user ID, please click "Login" to access the info.New users can click "Sign Up" to register for an user ID.

Preview

Preview

Codes

  • NII Article ID (NAID) :
    110007325722
  • NII NACSIS-CAT ID (NCID) :
    AN10013061
  • Text Lang :
    JPN
  • Article Type :
    Journal Article
  • ISSN :
    09135685
  • NDL Article ID :
    10203635
  • NDL Source Classification :
    ZN33(科学技術--電気工学・電気機械工業--電子工学・電気通信)
  • NDL Call No. :
    Z16-940
  • Databases :
    CJP  CJPref  NDL  NII-ELS 

Export