| Project/Area Number |
18K18025
|
|---|
| Research Category |
Grant-in-Aid for Early-Career Scientists
|
| Allocation Type | Multi-year Fund |
|---|
| Review Section |
Basic Section 60040:Computer system-related
|
|---|
| Research Institution | Nara Institute of Science and Technology |
|---|
Principal Investigator |
Shintani Michihiro 奈良先端科学技術大学院大学, 先端科学技術研究科, 助教 (80748913)
|
|---|
| Project Period (FY) |
2018-04-01 – 2021-03-31
|
| Project Status |
Completed (Fiscal Year 2020)
|
| Budget Amount *help |
¥3,510,000 (Direct Cost: ¥2,700,000、Indirect Cost: ¥810,000)
Fiscal Year 2020: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2019: ¥1,040,000 (Direct Cost: ¥800,000、Indirect Cost: ¥240,000)
Fiscal Year 2018: ¥1,430,000 (Direct Cost: ¥1,100,000、Indirect Cost: ¥330,000)
|
|---|
| Keywords | ニューラルネットワーク / メモリスタ / 高信頼化 / 誤り訂正 / 脳型コンピュータ / 過渡故障 / 永久故障 / 誤り訂正符号 / 機械学習 / 信頼性 / 検査容易化 |
|---|
| Outline of Final Research Achievements |
In this research, we conducted research on neuromorphic hardware that drastically improves the performance of deep learning, and in particular, we focused on improving the reliability of neural networks using memristors. Manufacturing technology of the memristors is immature and have issues with reliability. Therefore, we worked on developing the fault-tolerant techniques for the memoristar-based neural network. In particular, we have developed an error correction function that adds redundant cells for checksums in the column and row directions. As a result of numerical calculation using the Hopfield network, it was confirmed that the identification rate was improved by 25.81% compared to no measures and 5.25% compared to the existing method.
|
| Academic Significance and Societal Importance of the Research Achievements |
フォンノイマンボトルネックによるノイマン型計算基盤の性能向上の限界、ムーアの法則 の破綻が近づき、ヒトの脳を模した脳型コンピュータ(Neuromorphic Computer)はこれまでの計算の質を革新する最重要技術の1つとして期待を集めている。しかし、構成素子であるメムデバイスの製造不安定に起因する信頼性課題により大規模化、すなわち量、に重大な課題を抱えており、ノイマン型計算規模を超える目処は立っていない。本研究は、脳型コンピュータの高集積化に不可欠な高信頼化設計の技術基盤を形成するものであり、今後の脳型コンピュータの高集積化に向けた基礎技術となりうる。
|
