| 一类神经网络多层感知器的硬件实现方法 |
| |
| 引用本文： | 何述东,黄献青,瞿坦,黄心汉. 一类神经网络多层感知器的硬件实现方法[J]. 华中科技大学学报(自然科学版), 1996, 0(8) |
| |
| 作者姓名： | 何述东 黄献青 瞿坦 黄心汉 |
| |
| 作者单位： | He Shudong Dept. of Auto. Contr. Eng.，HUST，Wuhan 430074，China. Huang Xianqing Qu Tan Huang Xinhan |
| |
| 摘 要： | 提出了一种新的神经网络多层感知器硬件实现方法——信号传播路径动态选择．给出了隐节点个数的确定规则和网络的训练算法，并从理论上给予了证明．结果表明，该算法能一次确定权值，而且所确定的权值均为逻辑值，从而可以用逻辑门电路全硬件实现多层感知器
|
| 关 键 词： | 神经网络；多层感知器；硬件实现 |
Hardware Realization of the ANN Multilayer Perceptron |
| |
| He Shudong Dept. of Auto. Contr. Eng.,HUST,Wuhan ,China. Huang Xianqing Qu Tan Huang Xinhan. Hardware Realization of the ANN Multilayer Perceptron[J]. JOURNAL OF HUAZHONG UNIVERSITY OF SCIENCE AND TECHNOLOGY.NATURE SCIENCE, 1996, 0(8) |
| |
| Authors: | He Shudong Dept. of Auto. Contr. Eng. HUST Wuhan China. Huang Xianqing Qu Tan Huang Xinhan |
| |
| Affiliation: | He Shudong Dept. of Auto. Contr. Eng.,HUST,Wuhan 430074,China. Huang Xianqing Qu Tan Huang Xinhan |
| |
| Abstract: | The self learning function of the multilayer perceptron of an artificial neural network can be easily realized by the dynamic change of weights, but it is very difficult to achieve such a change in practice. A simple and quick training algorithm for accomplishing the learning process of t different training samples is proposed. The rule for determining the number of hidden nodes is also given. By the algorithm proposed, the weights can be obtained with only one iteration. The principle of the algorithm is proved and an example showing the details is given. |
| |
| Keywords: | neural network multilayer perceptron hardware realization |
| 本文献已被 CNKI 等数据库收录！ |
