Abstract:
Artificial neural networks (ANNs) are new technology emerged from approximate simulation of human brain and they have been successfully applied in many fields. Many researchers have tried to achieve optimal or near-optimal weights in artificial neural networks by using efficient methods. The traditional back propagation learning type requires huge number of training cycles and higher network configuration. Genetic algorithm (GA) can perform global search as against the local one performed by the gradient-based methods. Thus, GA can easily handle functions that are highly non-linear, complex, and noisy whereas the traditional gradient-based methods are inefficient. In this paper, genetic algorithm is used to adjust weight units which are important to improve network training in artificial neural networks. In the resulting ANNs-GA optimization approach, a trained ANN serves as an input-output model whose inputs are optimized by using the GA methodology. GA is used as embedded feature selection method to select relevant attributes before applying them to ANNs. And the proposed method diagnoses the medical datasets and compares accuracy of ANNs.
