{"title":"A New Approach to Polynomial Neural Networks based on Genetic Algorithm","authors":"S. Farzi","country":null,"institution":"","volume":20,"journal":"International Journal of Computer and Information Engineering","pagesStart":2700,"pagesEnd":2708,"ISSN":"1307-6892","URL":"https:\/\/publications.waset.org\/pdf\/3930","abstract":"Recently, a lot of attention has been devoted to\r\nadvanced techniques of system modeling. PNN(polynomial neural\r\nnetwork) is a GMDH-type algorithm (Group Method of Data\r\nHandling) which is one of the useful method for modeling nonlinear\r\nsystems but PNN performance depends strongly on the number of\r\ninput variables and the order of polynomial which are determined by\r\ntrial and error. In this paper, we introduce GPNN (genetic\r\npolynomial neural network) to improve the performance of PNN.\r\nGPNN determines the number of input variables and the order of all\r\nneurons with GA (genetic algorithm). We use GA to search between\r\nall possible values for the number of input variables and the order of\r\npolynomial. GPNN performance is obtained by two nonlinear\r\nsystems. the quadratic equation and the time series Dow Jones stock\r\nindex are two case studies for obtaining the GPNN performance.","references":null,"publisher":"World Academy of Science, Engineering and Technology","index":"Open Science Index 20, 2008"}