论文部分内容阅读
为研究图像和语音的模式分类,提出一种采用可变长度串遗传算法(VGA)的进化神经网络.该算法可以全局搜索优化神经网络的结构,找到神经网络接近最优的连接权,再通过反向传播算法(BP),在该优化结构中找到最优连接权.对语音数据和SPOT图像数据的验证结果表明,在模式分类中,采用该算法的分类器(VGA-BP)的分类性能较贝叶斯(Bayes)分类器、最近邻规则(k-NN)分类器具有更高的分类精度.
In order to study the pattern classification of images and speech, an evolutionary neural network based on variable length string genetic algorithm (VGA) is proposed, which can globally search and optimize the structure of neural network to find the optimal near-right neural network connection, Backpropagation algorithm (BP) is used to find the optimal connection weight in the optimized structure.The verification results of the speech data and the SPOT image data show that in the pattern classification, the classification performance of the classifier (VGA-BP) Compared with Bayes classifier, nearest neighbor rule (k-NN) classifier has higher classification accuracy.