A Combine Algorithm for a Cmac Network

The performance of a CMAC neural network depends on the training algorithms and the selection of input points. Papers have been published that explain CMAC algorithms but little work has been done to improve existing algorithms. In this paper, the existing algorithms are first explained and then compared using computational results and the algorithm properties. Improvements are made to the recommended Maximum Error Algorithm by using a "Combine Algorithm" approach. In this method, CMAC network is first trained by using Neighborhood Training Algorithm and then trained by Maximum Error Algorithm for fine-tuning of CMAC network. Faster initial convergence is achieved for the recommended Maximum Error Algorithm. This approach may reduce the training time and accelerate the initial learning which is very important in many control applications.