2022 Fiscal Year Final Research Report
Declarative memory model between hippocampus and neocortex
| Project/Area Number |
19K12155
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Review Section |
Basic Section 61040:Soft computing-related
|
|---|
| Research Institution | University of Yamanashi |
|---|
Principal Investigator |
|
|---|
| Project Period (FY) |
2019-04-01 – 2023-03-31
|
| Keywords | 海馬 / 大脳皮質 / エピソード記憶 / 宣言的記憶 / ニューラルネットワーク / 破局的忘却 / スパイキングニューラルネットワーク / 生成モデル |
|---|
| Outline of Final Research Achievements |
In human memory, memories related to facts and events are called declarative memories and are the basis of our intelligent information processing in the brain. In this study, we aimed to construct an intelligent system that mimics the brain and is useful from an engineering perspective. We modeled the formation process of declarative memory in the brain and obtained the following results: (1) We devised a new supervised learning method for spiking neural networks, which use elaborate models of neurons, and demonstrated the possibility of memory formation with suppression of catastrophic forgetting. (2) We showed that a new hippocampal model introducing the CA2 region enables the storage and retrieval of complex episodes. (3) We investigated the ease of memorization and forgetfulness of information and devised an effective learning method to suppress forgetting. (4) We developed learning methods to improve long-term memory robustness and generalization performance.
|
| Free Research Field |
情報工学
|
| Academic Significance and Societal Importance of the Research Achievements |
意識的に思い出すことのできる記憶(宣言的記憶)は，思考や推論といった脳の高次情報処理に利用されているため，人間のように知的で柔軟な情報処理システムを構築するにあたっては，如何にして宣言的記憶の形成過程を模倣するかが重要である．特に人間特有の記憶については，脳の一部を損傷させて機能を解明するような研究は倫理的に不可能なため，計算モデルによる研究の果たす役割が大きい．本研究では，主に脳の海馬に関する知見に基づき，宣言的記憶を形成するモデルの構築，忘却を抑制しつつ記憶を形成するための学習方法の構築を行い，人間の記憶の形成過程を工学的に模倣するモデルの基礎としての成果を得た．
|
