粗糙集意义下的一种RBF神经网络设计方法

提出一种集成粗糙集理论的RBF网络设计方法.由布尔逻辑推理方法进行属性离散化,得到初始决策模式集,通过差异度对初始决策模式的相似度进行衡量并实现聚类,以聚类决策模式构造RBF网络.为加快训练速度,分别对隐层参数和输出权值采用BP算法和线性最小二乘滤波法进行训练.实验结果表明,该方法设计的RBF网络结构简洁,泛化性能良好,混合学习算法的收敛速度优于单纯的BP算法.

A method of designing RBF neural network based on rough sets

A method of designing RBF neural network, which integrates rough sets theory, is proposed. Continuous attributes are discretized by using Boolean reasoning algorithm and original decision modes are extracted. Similarities among original decision modes can be measured by dissimilarity degree and original decision modes can be clustered. Clustered decision modes are used to construct RBF neural network. To increase the training speed, a hybrid training algorithm is introduced, in which the parameters of hidden layer and weights of output layer are tuned by using back propagation algorithm and linear least squares filtering, respectively. Experiment results show that the designed RBF neural network has refined structure and good generalization ability. The convergence speed of hybrid training algorithm is superior to the single back propagation algorithm.