| Project/Area Number |
15H01706
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (A)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Soft computing
|
|---|
| Research Institution | Kyushu Institute of Technology |
|---|
Principal Investigator |
Morie Takashi 九州工業大学, 大学院生命体工学研究科, 教授 (20294530)
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
高橋 庸夫 北海道大学, 情報科学研究科, 教授 (90374610)
寒川 誠二 東北大学, 流体科学研究所, 教授 (30323108)
遠藤 和彦 国立研究開発法人産業技術総合研究所, エレクトロニクス・製造領域, 研究グループ長 (60392594)
|
|---|
| Research Collaborator |
TAMUKOH Hakaru
OHNO Takeo
KUBOTA Tomohiro
TOHARA Takashi
ANDO Hideyuki
TOMIZAKI KAZUMASA
KATO Takashi
WANG Quan
TANIMURA Hiroshi
YAMAGUCHI Masatoshi
HARADA Masataka
IWAMOTO Goki
YAMASHITA Kenya
|
|---|
| Project Period (FY) |
2015-04-01 – 2019-03-31
|
| Project Status |
Completed (Fiscal Year 2018)
|
| Budget Amount *help |
¥43,420,000 (Direct Cost: ¥33,400,000、Indirect Cost: ¥10,020,000)
Fiscal Year 2018: ¥8,710,000 (Direct Cost: ¥6,700,000、Indirect Cost: ¥2,010,000)
Fiscal Year 2017: ¥11,050,000 (Direct Cost: ¥8,500,000、Indirect Cost: ¥2,550,000)
Fiscal Year 2016: ¥11,440,000 (Direct Cost: ¥8,800,000、Indirect Cost: ¥2,640,000)
Fiscal Year 2015: ¥12,220,000 (Direct Cost: ¥9,400,000、Indirect Cost: ¥2,820,000)
|
|---|
| Keywords | ソフトコンピューティング / ニューラルネットワーク / 脳型人工知能ハードウェア / 抵抗変化型メモリ素子 / ナノ構造 / 集積回路 |
|---|
| Outline of Final Research Achievements |
To realize highly integrated and highly power efficient hardware implementing brain-like intelligent processing, we proposed a time-domain analog circuit architecture, and developed a large-scale integrated circuit (LSI) that achieves 30 times higher efficiency (lower energy consumption) using ten times larger fabrication technology compered to state-of-the-art digital LSIs. If the latest fabrication technology is used, more than 100 times higher efficiency can be achieved. As memory devices required to this circuit architecture, we developed fabrication technology for resistance change random access memory (ReRAM) devices that realize analog memory property and high resistance. We also proposed a parallel connected ReRAM-MOSFET device structure for stable analog operation and evaluated its function. We also fabricated nanostructures generating noise, which is required for brain-like processing, and their effectiveness was evaluated.
|
| Academic Significance and Societal Importance of the Research Achievements |
社会に普及が進んでいる人工知能（AI）をさらに高度化し，人の脳の機能に近づけるために，脳型処理モデルとそのハードウェア化の研究が必要である．本研究では，現在のデジタル計算機主体のAIに対して飛躍的に高効率化できる新しい集積回路方式を提案し，それに必要なデバイス，特にアナログメモリ素子とノイズ発生素子についての基盤技術を開発した．この技術を発展させることで，人々に寄り添うAIの開発が進展することが期待できる．
|
