混沌遗传算法对BP神经网络的改进研究

BP神经网络算法是目前应用最广泛的一种神经网络算法,但有收敛速度慢和易陷入局部极小值等缺陷.本文利用混沌遗传算法(CGA)具有混沌运动遍历性、遗传算法反演性的特性来改进BP神经网络算法.该算法的基本思想是用混沌遗传算法对BP神经网络算法的初始权值和初始阈值进行优化.把混沌变量加入遗传算法中,提高遗传算法的全局搜索能力和收敛速度;用混沌遗传算法优化后得到的最优解作为BP神经网络算法的初始权值和阈值.通过实验观察,改进后的结果与普通的BP神经网络算法的结果相比,具有更高的准确率.

An Improvement to the BP Neural Network Algorithm Based on the Chaos Genetic Algorithm

The BP neural network algorithm is the most common neural network algorithm with a wide range of practi- cal applications. However it has several defects such as slow convergence speed and easiness to getting stuck into lo- cal optima. In this paper by making use of the ergodicity and recurrence of the chaotic genetic algorithm （ CGA ） , a new algorithm is proposed to improve the BP neural network algorithm. The basic idea is to optimize the initial weights and threshold of the BP network algorithm and to add chaotic variables to improve the global searching ability and con- vergence speed. The optimal resuhs of the chaotic genetic algorithm are used as the initial weights and threshold of the BP network algorithm. The effectiveness of our improved BP neural network algorithm is demonstrated by some nu- merical examples.