一种有效的最优序参量重构方法  

An Effective Method of Optimal Reconstruction of Order Parameters

作　　者：王海龙[1] 戚飞虎[1] 

机构地区：[1]上海交通大学计算机科学与工程系,上海200030

出　　处：《中国图象图形学报（A辑）》2001年第1期56-60,共5页Journal of Image and Graphics

基　　金：国家自然科学基金!资助项目 (6 9772 0 0 2 )

摘　　要：针对模式识别中协同方法存在的问题 ,提出了一种协同神经网络中序参量重构的方法 ,该方法是利用遗传算法的全局最优搜索能力 ,通过对训练样本集的学习 ,然后再通过在序参量的重构参数空间进行全局搜索来获得最优重构参数 .利用实际采样得到的样本对新算法进行的测试表明 ,新方法确实能找到一组序参量重构参数 ,并能使识别性能有较大提高 .A novel method of reconstruction of order parameters in synergetic neural network is presented in this paper, Considering the neural network has the self learning ability, so we construct this linear transformantion using this ability. Additionally, considering the genetic algorithm has the globally optimal searching ability, so we can achieve these reconstruction parameters using genetic algorithm. The new method trained the synergetic neural network using genetic algorithm on the training samples set, after the convergence of genetic algorithm, the reconstruction parameters can be got. In the theory, genetic algorithm can achieve the globally optimal reconstruction parameters after infinite computation, which can be guaranteed by itself theory. So this method completely solves the construction of reconstruction parameters on the theory. The test on the samples from real applications shows:new method really can find a group of reconstruction parameters which improves the performance of synergetic neural network greatly.

关 键 词：序参量 序参量重构 协同神经网络 遗传算法 模式识别 

分 类 号：TP391.4[自动化与计算机技术—计算机应用技术]