An Adaptive Type Comeptitive Network for Pattern seqmentation in associative Memory

  • Ba Sei
    Communications Research Laborotory,MPT

Bibliographic Information

Other Title
  • 連想記憶におけるパターン分離の適応型競合回路

Search this article

Description

In associative memory,pattern segmentation is an essential and indispensable function for representing(storing and recalling)one- to-multiple associations,or,for complementing patterns from their partials common to different original patterns.This report presents an adaptive type competitive network having high performance for pattern segmentation,by introducing an adaptive type of excitatory connection in addition to the tearnable type of inhibitory one used in the HASP type competitive network.The results of computer simulation show that the new competitive network is superior in both separation ability and speed of pattern segmentation to the previous networks.

Journal

Keywords

Details 詳細情報について

  • CRID
    1574231877214212352
  • NII Article ID
    110003233230
  • NII Book ID
    AN10091178
  • Text Lang
    ja
  • Data Source
    • CiNii Articles

Report a problem

Back to top