Inverse Function Delayed Model for Optimization Problems
Inverse Function Delayed Model for Optimization Problems
复制标题
DOI：
10.1007/978-3-540-30132-5_132
复制
发表时间：
2004-09
期刊：
影响因子：
--
通讯作者：
Y. Hayakawa;Tatsuaki Denda;K. Nakajima
中科院分区：
文献类型：
--
作者：
Y. Hayakawa;Tatsuaki Denda;K. Nakajima
ID模型是由附着在常规网络作用上的宏观模型衍生而来的一种新模型，其重要特点是可以引入负阻力效应。本文利用负阻力效应的作用，解决了局部最小状态的不稳定问题，这是神经网络优化问题的一大难题，并通过数值实验证明了该方法的良好性能。
The ID model is a novel model derived from a macroscopic model that is attached to conventional network action, and the important character is what we can introduce negative resistance effect into. In this paper, we aim at the unstabilization of local minimum states, which is a big problem to solving optimization problems in a neural network, by the action of this negative resistance effect, and we show the good performance by numerical experiments.