2013 Fiscal Year Final Research Report
High dimensional neural networks using simulations perturbation learning rule and their hardware implementation
| Project/Area Number |
23500290
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (C)
|
| Allocation Type | Multi-year Fund |
|---|
| Section | 一般 |
|---|
| Research Field |
Sensitivity informatics/Soft computing
|
|---|
| Research Institution | Kansai University |
|---|
Principal Investigator |
MAEDA Yutaka 関西大学, システム理工学部, 教授 (60209393)
|
|---|
| Project Period (FY) |
2011 – 2013
|
| Keywords | ニューラルネットワーク / 同時摂動 / クリフォード代数 / ハードウェア / ＦＰＧＡ / 制御 |
|---|
| Research Abstract |
In this research, we handled the following points. 1) We proposed learning schemes for high dimensional neural networks based on the simulations perturbation optimization method. We confirmed that the proposed schemes have good performance equal to the ordinary back-propagation method. 2) We considered hardware pulse density complex-valued neural network with the simulations perturbation learning rule based on FPGA system. We made the system experimentally. The system leant some basic benchmark problems properly. 3) As an application of the proposed high dimensional neural networks, we handled control problems for robot systems. The neural networks learnt inverse kinematics of objective robots and have generalization capability.
|
