Development of High-speed Information Processing System and Its Application to the Bioengineering Signal Processing
Project/Area Number |
02680029
|
Research Category |
Grant-in-Aid for General Scientific Research (C)
|
Allocation Type | Single-year Grants |
Research Field |
Informatics
|
Research Institution | The University of Tokushima |
Principal Investigator |
AKAMATSU Norio Tokushima University,Faculty of Engineering,Professor, 工学部, 教授 (20035629)
|
Co-Investigator(Kenkyū-buntansha) |
河辺 徹 徳島大学, 工学部, 助手 (40224844)
|
Project Period (FY) |
1990 – 1992
|
Project Status |
Completed (Fiscal Year 1992)
|
Budget Amount *help |
¥1,400,000 (Direct Cost: ¥1,400,000)
Fiscal Year 1992: ¥400,000 (Direct Cost: ¥400,000)
Fiscal Year 1991: ¥400,000 (Direct Cost: ¥400,000)
Fiscal Year 1990: ¥600,000 (Direct Cost: ¥600,000)
|
Keywords | Neural computer / Surface maps / Pattern recognition / Connectionism / Neural computer architecture / Connection weights / High-speed information processing system / Large-scale integrated circuits / マルチセンサ- / ニュ-ラル・コンピュ-タ / 電磁ニュ-ラル・コンピュ-タ / パタ-ン分類 / 体表面電位図 / マルチプロセッサ / ニュ-ロネットワ-ク / 並列情報処理 / ニュ-ロン / 電磁結合 |
Research Abstract |
In order to realize a high-speed information processing system,we have developed its hardware and software. On the basis of the measured data,we focus our attention to the feature extraction of body surface maps according to the difference of age and sex so as to accomplish pattern recognition of heart. A neural computer is referred to as a connectionism or an ultra-parallel distributed processing system,compared to a conventional computer.However,it is necessary to develop an effective wire connecting method among neurons to realize a large-scale network involving a huge number of neurons. This project proposes a new neural computer architecture in which each layer of the network is coupled electro-magnetically. The effectiveness of the new neural computer is confirmed by computer simulation. Since the connection weights in the neural computer can be realized without the need for special devices,the proposed neural computer may easily be produced by the technology used for large-scale integrated circuits.
|
Report
(4 results)
Research Products
(17 results)