Accumulator based arbitration model for both supervised and reinforcement learning inspired by prefrontal cortex

Masahiko Osawa, Yuta Ashihara, Takuma Seno, Michita Imai, Satoshi Kurihara

Research output: Chapter in Book/Report/Conference proceedingConference contribution

4 Citations (Scopus)

Abstract

A method that provides an excellent performance by arbitrating multiple modules is important. There are variety of multi-module arbitration methods proposed in various contexts. However, there is yet to be a multi-module arbitration method proposed in reference to structure of animals’ brains. Considering that the animals’ brains achieve general-purpose multi-module arbitration, such function may be achieved by referring to the actual brain. In this paper, with reference to the knowledge of accumulator neurons hypothesized to exist in the prefrontal cortex, we propose an Accumulator Based Arbitration Model (ABAM). By arbitrating multiple modules, ABAM exerts a superior performance in both supervised learning and reinforcement learning task.

Original languageEnglish
Title of host publicationNeural Information Processing - 24th International Conference, ICONIP 2017, Proceedings
EditorsYuanqing Li, Derong Liu, Shengli Xie, El-Sayed M. El-Alfy, Dongbin Zhao
PublisherSpringer Verlag
Pages608-617
Number of pages10
ISBN (Print)9783319700861
DOIs
Publication statusPublished - 2017
Event24th International Conference on Neural Information Processing, ICONIP 2017 - Guangzhou, China
Duration: 2017 Nov 142017 Nov 18

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10634 LNCS
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other24th International Conference on Neural Information Processing, ICONIP 2017
Country/TerritoryChina
CityGuangzhou
Period17/11/1417/11/18

Keywords

  • Accumulator model
  • Ensemble learning
  • Hierarchical architecture
  • Prefrontal cortex

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Fingerprint

Dive into the research topics of 'Accumulator based arbitration model for both supervised and reinforcement learning inspired by prefrontal cortex'. Together they form a unique fingerprint.

Cite this