基于遗传优化的粗糙神经网络模式识别器及应用

A Rough Neural Network Pattern Recognition Apparatus Based on Genetic Optimization and Its Application

通过粗糙集获取知识表达系统的分类规则,用产生的规则对神经网络进行编码,并利用遗传算法对初步学习后的神经网络的权值进行优化,最终得到一个神经网络模式识别器。举例说明了采用这种方法得到模式识别器的过程及其对待识别对象的学习和分类效果。结果表明:采用粗糙规则对神经网络编码可以缩短神经网络的训练过程,遗传算化对神经网络权值的优化可在一定程度上提高模式识别的精度。

Gives a pattern recognition apparatus based on picking -up rules of recognizable objects and u-sing these rules to encode and train the neural network. Uses genetic algorithm to optimize the networks ' weights. Gives an example to show the pattern recognition apparatus has high efficiency in pattern recognition. The result of the example shows the encoding of rules can shorten the time of net - training and the optimization of the net weights can improve the pression of recognition.