An implementation of working memory using stacked half restricted Boltzmann machine: Toward to restricted Boltzmann machine-based cognitive architecture

Masahiko Osawa, Hiroshi Yamakawa, Michita Imai

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

1 Citation (Scopus)

Abstract

Cognition, judgment, action, and expression acquisition have been widely treated in studies on recently developed deep learning. However, although each study has been specialised for specific tasks and goals, cognitive architecture that integrates many different functions remains necessary for the realisation of artificial general intelligence. To that end, a cognitive architecture fully described with restricted Boltzmann machines (RBMs) in a unified way are promising, and we have begun to implement various cognitive functions with an RBM base. In this paper, we propose new stacked half RBMs (SHRBMs) made from layered half RBMs (HRBMs) that handle working memory. We show that an ability to solve maze problems that requires working memory improves drastically when SHRBMs in the agent’s judgment area are used instead of HRBMs or other RBM-based models.

Original languageEnglish
Title of host publicationNeural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings
PublisherSpringer Verlag
Pages342-350
Number of pages9
Volume9947 LNCS
ISBN (Print)9783319466866
DOIs
Publication statusPublished - 2016
Event23rd International Conference on Neural Information Processing, ICONIP 2016 - Kyoto, Japan
Duration: 2016 Oct 162016 Oct 21

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume9947 LNCS
ISSN (Print)03029743
ISSN (Electronic)16113349

Other

Other23rd International Conference on Neural Information Processing, ICONIP 2016
CountryJapan
CityKyoto
Period16/10/1616/10/21

Fingerprint

Boltzmann Machine
Cognitive Architecture
Working Memory
Data storage equipment
Cognition
Integrate
Necessary
Deep learning

Keywords

  • Cognitive architecture
  • Restricted Boltzmann machine

ASJC Scopus subject areas

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Osawa, M., Yamakawa, H., & Imai, M. (2016). An implementation of working memory using stacked half restricted Boltzmann machine: Toward to restricted Boltzmann machine-based cognitive architecture. In Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings (Vol. 9947 LNCS, pp. 342-350). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9947 LNCS). Springer Verlag. https://doi.org/10.1007/978-3-319-46687-3_38

An implementation of working memory using stacked half restricted Boltzmann machine : Toward to restricted Boltzmann machine-based cognitive architecture. / Osawa, Masahiko; Yamakawa, Hiroshi; Imai, Michita.

Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings. Vol. 9947 LNCS Springer Verlag, 2016. p. 342-350 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 9947 LNCS).

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

Osawa, M, Yamakawa, H & Imai, M 2016, An implementation of working memory using stacked half restricted Boltzmann machine: Toward to restricted Boltzmann machine-based cognitive architecture. in Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings. vol. 9947 LNCS, Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 9947 LNCS, Springer Verlag, pp. 342-350, 23rd International Conference on Neural Information Processing, ICONIP 2016, Kyoto, Japan, 16/10/16. https://doi.org/10.1007/978-3-319-46687-3_38
Osawa M, Yamakawa H, Imai M. An implementation of working memory using stacked half restricted Boltzmann machine: Toward to restricted Boltzmann machine-based cognitive architecture. In Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings. Vol. 9947 LNCS. Springer Verlag. 2016. p. 342-350. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-46687-3_38
Osawa, Masahiko ; Yamakawa, Hiroshi ; Imai, Michita. / An implementation of working memory using stacked half restricted Boltzmann machine : Toward to restricted Boltzmann machine-based cognitive architecture. Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings. Vol. 9947 LNCS Springer Verlag, 2016. pp. 342-350 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{d01d2224e3994d37948a9d18801c83c1,
title = "An implementation of working memory using stacked half restricted Boltzmann machine: Toward to restricted Boltzmann machine-based cognitive architecture",
abstract = "Cognition, judgment, action, and expression acquisition have been widely treated in studies on recently developed deep learning. However, although each study has been specialised for specific tasks and goals, cognitive architecture that integrates many different functions remains necessary for the realisation of artificial general intelligence. To that end, a cognitive architecture fully described with restricted Boltzmann machines (RBMs) in a unified way are promising, and we have begun to implement various cognitive functions with an RBM base. In this paper, we propose new stacked half RBMs (SHRBMs) made from layered half RBMs (HRBMs) that handle working memory. We show that an ability to solve maze problems that requires working memory improves drastically when SHRBMs in the agent’s judgment area are used instead of HRBMs or other RBM-based models.",
keywords = "Cognitive architecture, Restricted Boltzmann machine",
author = "Masahiko Osawa and Hiroshi Yamakawa and Michita Imai",
year = "2016",
doi = "10.1007/978-3-319-46687-3_38",
language = "English",
isbn = "9783319466866",
volume = "9947 LNCS",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "342--350",
booktitle = "Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings",
address = "Germany",

}

TY - GEN

T1 - An implementation of working memory using stacked half restricted Boltzmann machine

T2 - Toward to restricted Boltzmann machine-based cognitive architecture

AU - Osawa, Masahiko

AU - Yamakawa, Hiroshi

AU - Imai, Michita

PY - 2016

Y1 - 2016

N2 - Cognition, judgment, action, and expression acquisition have been widely treated in studies on recently developed deep learning. However, although each study has been specialised for specific tasks and goals, cognitive architecture that integrates many different functions remains necessary for the realisation of artificial general intelligence. To that end, a cognitive architecture fully described with restricted Boltzmann machines (RBMs) in a unified way are promising, and we have begun to implement various cognitive functions with an RBM base. In this paper, we propose new stacked half RBMs (SHRBMs) made from layered half RBMs (HRBMs) that handle working memory. We show that an ability to solve maze problems that requires working memory improves drastically when SHRBMs in the agent’s judgment area are used instead of HRBMs or other RBM-based models.

AB - Cognition, judgment, action, and expression acquisition have been widely treated in studies on recently developed deep learning. However, although each study has been specialised for specific tasks and goals, cognitive architecture that integrates many different functions remains necessary for the realisation of artificial general intelligence. To that end, a cognitive architecture fully described with restricted Boltzmann machines (RBMs) in a unified way are promising, and we have begun to implement various cognitive functions with an RBM base. In this paper, we propose new stacked half RBMs (SHRBMs) made from layered half RBMs (HRBMs) that handle working memory. We show that an ability to solve maze problems that requires working memory improves drastically when SHRBMs in the agent’s judgment area are used instead of HRBMs or other RBM-based models.

KW - Cognitive architecture

KW - Restricted Boltzmann machine

UR - http://www.scopus.com/inward/record.url?scp=84992561434&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84992561434&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-46687-3_38

DO - 10.1007/978-3-319-46687-3_38

M3 - Conference contribution

AN - SCOPUS:84992561434

SN - 9783319466866

VL - 9947 LNCS

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 342

EP - 350

BT - Neural Information Processing - 23rd International Conference, ICONIP 2016, Proceedings

PB - Springer Verlag

ER -