摘要
根据遗传算法的演化原理,提出了一种新的度量模糊Hamming网络参数性能优劣的适应度函数,在此基础上,将遗传算法作为模糊Hamming网络的训练算法,进行网络参数寻优,并采用UCI机器学习数据库中不同的数据集进行对比测试。试验结果表明:将遗传算法与模糊Hamming网络相结合,可以快速求得最优网络参数组合,避免原有参数调整时的盲目性和随机性,训练后的网络具有分类迅速、准确、稳定的特点。
According to the evolutionary principle of GA,a new fitness function for the fuzzy hamming net is proposed,which can measure the performance of different combinations of the key parameters of that net.Then the genetic algo-rithm is applied to training the net to find the optimal combination of these parameters.And the datasets from UCI ma-chine learning database are used to examine the training algorithm proposed.Experimental evidence shows that the blindness and randomicity of the manual adjustment ,which is previously used with this net,can be eliminated,the satis-factory solution can be found quickly,and the trained net possessed the characteristic of quickness,correctness and sta-bleness during the classification.
出处
《计算机工程与应用》
CSCD
北大核心
2004年第20期11-13,共3页
Computer Engineering and Applications
基金
国家自然科学基金项目:基于测度论的系统评价公理化研究(编号:70371075)资助
