基于模糊Hamming网络的模式识别的研究

A Research on the Pattern Recognition Based on Fuzzy Hamming Net

探讨了一种模糊Ｈａｍｍｉｎｇ网络，它通过设置阈值对隐含层的神经元进行择优来代替模糊ＡＲＴ网络的搜索循环过程，克服了模糊ＡＲＴ网络在处理大样本问题时搜索时间长的问题，同时保持了模糊ＡＲＴ的良好性能．对该网络进行了模式识别实验，并与模糊ＡＲＴ进行了比较性试验，结果表明，该网络具有良好的分类能力和较高的学习效率．

A fuzzy Hamming network is discussed in this paper. By setti ng a threshold to the hidden layer，the best neutrons are selected. Wh ich replace the search cycle and overcome the problem of long search t ime in dealing with large samples in fuzzy ART net. At the same time i t remains the good properties of fuzzy ART net. An experiment is carri ed out on the pattern recognition of the lithergy samples. And the res ults are compared with the fuzzy ART net. It shows that the fuzzy Hamm ing net has a better classification probability and a higher learning rate.