A Neural Network Solver for Differential Equations

摘要：

We propose a solver for differential equations, which only use one neural network, building of multi-layer-structure and can be learned. The learning method is defined as an equation that resembles the BP. The techniques are based on the analogue type of neural network, its derivative expression and iterations which is similar to the BP algorithm. Precision of the solution depends on the learning-error of the analogue type of neural network. The structure of the solver is equivalent to the multi-layer neural network; therefore, a parallel processing can be done.

关键词： neural network differential equation derivative of network

作者: Qianyi WANG Tomoo AOYAMA

作者单位: The faculty of engineering, Miyazaki University Gakuen Kibanadai-nishi 1-1, Miyazaki city, 889-2192, Japan

会议类型: 国际会议

会议名称: 8th International Conference on Neural Information Processing(ICONIP 2001)(第八届国际神经信息处理大会)

会议地点: 上海

会议语种:英文

页码: 1117-1120

在线出版日期: 2001-11-14（万方平台首次上网日期，不代表论文的发表时间）