• Search Research Projects
  • Search Researchers
  • How to Use
  1. Back to project page

2021 Fiscal Year Final Research Report

Study on neural network model of medial temporal lobe as brain of a rat type robot

Research Project

  • PDF
Project/Area Number 17K00344
Research Category

Grant-in-Aid for Scientific Research (C)

Allocation TypeMulti-year Fund
Section一般
Research Field Soft computing
Research InstitutionKyushu Institute of Technology

Principal Investigator

Tateno Katsumi  九州工業大学, 大学院生命体工学研究科, 准教授 (00346868)

Project Period (FY) 2017-04-01 – 2022-03-31
Keywordsスパイキングニューラルネットワーク / 内側側頭葉 / 嗅内皮質 / 海馬 / GPU
Outline of Final Research Achievements

A library for parallel computation of spiking neural networks (SNN) has been prepared to provide an environment for high-speed computation of large-scale SNNs. Using this library, we created grid cells and head orientation cells in the entorhinal cortex and constructed SNNs that form place cells in the hippocampus. Adding an action selection SNN to the hippocampal SNN would allow for reward-dependent spatial learning. A SNN containing time cells was also created and spatial learning dependent on past pathways. With the high-speed computation now possible, connecting a mobile robot to the hippocampal SNN, place cells were formed in the hippocampal SNN in real-time based on the movement speed and head direction of the mobile robot. Time cell-like behavior of compartmental pyramidal cells in a SNN was also reproduced.

Free Research Field

計算論的神経科学

Academic Significance and Societal Importance of the Research Achievements

ニューラルネットワークの中でも、SNNを用いた研究が盛んに行われるようになってきた。誤差逆伝搬法のような人工ニューラルネットワークの理論がSNNに適用できるようになってきたことが理由である。一方で、SNNの計算は、連立微分方程式を数値解法により解くため、計算負荷が高い。本研究成果によるライブラリはGraphics Processing Unitを意識することなく、並列計算によりSNNを作成でき、高速な計算を容易にする点で意義がある。本研究は、海馬に特化し、空間の場所表現に関するSNNを対象としたが、別の脳領域のSNNの構築も可能であるので、より大規模で機能的なSNNを構築することを可能にする。

URL: 

Published: 2023-01-30  

Information User Guide FAQ News Terms of Use Attribution of KAKENHI

Powered by NII kakenhi