| Project/Area Number |
19K16885
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 51010:Basic brain sciences-related
|
|---|
| Research Institution | Okinawa Institute of Science and Technology Graduate University |
|---|
Principal Investigator |
FUNG CHI CHUNG 沖縄科学技術大学院大学, 神経情報・脳計算ユニット, 客員研究員 (80757203)
|
|---|
| Project Period (FY) |
2019-04-01 – 2024-03-31
|
| Project Status |
Completed (Fiscal Year 2023)
|
| Budget Amount *help |
¥4,290,000 (Direct Cost: ¥3,300,000、Indirect Cost: ¥990,000)
Fiscal Year 2022: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2021: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2020: ¥650,000 (Direct Cost: ¥500,000、Indirect Cost: ¥150,000)
Fiscal Year 2019: ¥2,340,000 (Direct Cost: ¥1,800,000、Indirect Cost: ¥540,000)
|
|---|
| Keywords | Hebbian Learning / Machine Learning / Adult Neurogenesis / Pattern Separation / Dentate Gyrus / Neuroscience / Unsupervised Learning / Synaptic Comeptition / Computational Model / Hebbian-like Plasticity / Memory Formation / Synaptic Plasticity / Detate Gyrus / Hippocampus / Slow Oscillation |
|---|
| Outline of Research at the Start |
Adult neurogenesis is a phenomenon that new neurons are generated in the dentate gyrus during adulthoods. Its impairment will suppress the formation of memory. However, how the maturation of those new cells can contribute to long-term memory formation is mostly unknown. In this study, the following issues will be addressed: (1) how the maturation schedule of new-born cells can alter the information processing performance; (2) the optimal learning strategy for information processing; and (3) how other brain regions (e.g., CA3) can make use of the optimal representation for different tasks.
|
| Outline of Final Research Achievements |
In this work, we found that synaptic competition improves the discrimination of overlapping memory patterns by reducing memory interference through competition between newly generated neurons and mature neurons. Neural networks trained with synaptic competition-based learning rules show superior performance, especially with fewer training samples, compared to those trained with backpropagation algorithms. Computational models indicate that synaptic competition plays a crucial role in information processing. Behavioral experiments in the literature confirm that mice with suppressed adult neurogenesis have difficulty discriminating interfering memories, supporting the importance of synaptic competition and neurogenesis.
|
| Academic Significance and Societal Importance of the Research Achievements |
本研究は、シナプス競争が記憶パターンの識別を改善し、成人期の神経新生が記憶の干渉を減少させるメカニズムを解明しました。シナプス競争に基づく学習ルールが少数の訓練サンプルで優れた性能を示し、情報処理における重要性を示す成果です。または、この研究は、アルツハイマー病などの記憶障害の新たな治療法の開発に寄与し、シナプス競争に基づく学習アルゴリズムはAIや機械学習の技術革新を促進する可能性があります。
|
