Research for a System that Recognize the Outer World Dynamically Using the Method of Neural Information Processing and Artificial Intelligence.
Project/Area Number |
60460126
|
Research Category |
Grant-in-Aid for General Scientific Research (B)
|
Allocation Type | Single-year Grants |
Research Field |
電子通信系統工学
|
Research Institution | University of Tokyo |
Principal Investigator |
AMARI Shun-ichi Fuculty of Eng., Univ. Tokyo, Prof., 工学部計数工学科, 教授 (80010726)
|
Co-Investigator(Kenkyū-buntansha) |
KURATA Koji Fuculty of Eng., Univ. Tokyo, Res. Assoc., 工学部計数工学科, 助手 (40170071)
MAGINU Kenjiro Kyuhsu Institute of Technology, Prfof., 情報工学科, 教授 (40011134)
|
Project Period (FY) |
1985 – 1987
|
Project Status |
Completed (Fiscal Year 1987)
|
Budget Amount *help |
¥5,500,000 (Direct Cost: ¥5,500,000)
Fiscal Year 1987: ¥1,500,000 (Direct Cost: ¥1,500,000)
Fiscal Year 1986: ¥2,500,000 (Direct Cost: ¥2,500,000)
Fiscal Year 1985: ¥1,500,000 (Direct Cost: ¥1,500,000)
|
Keywords | Reaction-Deffusion Equation / Stability / Nerve field / Topography / Columnar Structure / Associative Memory / Boltzmann Machine / 画像処理 / 並列情報処理 / 自己組織機能 / 画像の不変特徴 / 運動情報と形態情報の分離 / ボルツマンマシン |
Research Abstract |
The Brain is the best computer that we know. If we can understand and simulate the principle of its information processing, we will be able to make a computer that is completely different from those we have now. This is the aim of our project. It is classified into the following three parts. 1. The research of a single neuron. The nature of the pulse trains that travl on nerve fiber is studied in terms of the reaction-diffusion equation. A new topological characteristic is defined, and it is shown to have a close relation with the stability of the pulse trains. 2. The research of neural networks. A special kind of nerve field is studied that has the ability to learn the topology of the set of input signals. Under some condition the continuous mapping between the set of input signals and the field is unstable. in which case, it is shown, a kind of block structure is formed that reminds us of the columnar structure in the brain. Boltzmann machine, which is a thermo-dynamocal model of neuron networks, is also stukied theoretically. 3. The research of algorithms. A new parallel processing algorithm is proposed to extract the orientation and the motion of a surface.
|
Report
(2 results)
Research Products
(18 results)