Improved Cerebellum Controller Neural Network Algorithm
Pan Cheng-en · Jisuanji fangzhen · 2011
Study the cerebellum controller neural network(CMAC) in the application of pattern recognition.There are a lot of High dimension and redundant information in present neural network training samples,which often results in complex neural network structure,slow training,and low recognition rate.In this paper,a new pattern recognition model is proposed based on rough set and Cerebellar Model Articulation Controller(CMAC),which uses rough sets to reduce the redundant information and dimension of training sample to simplify the structure of neural networ,and a dynamic learning rate is introduced to accelerate the convergence speed and recognition rate.Simulation results show that this method can effectively improve the speed and accuracy compared with traditional CMAC models.RS_CMAC can overcome the shortcomings of traditional CMAC models,and is an effective pattern recognition method.