Meta-heuristic algorithms have been successfully used in hard continuous optimization problems. The training of Artificial Neural Networks (ANNs) is one of these hard continuous optimization problems which has been solved in the literature using different optimization algorithms. In this study, a Bat algorithm (BA) and a Genetic Algorithm (GA) is proposed for the training of the ANNs. The performance of the algorithms has been tested with well-known seven datasets from UCI (University of California, Irvine) machine learning repository. The obtained results are compared with Back-propagation (BP) learning algorithm. It is figured out that the ANNs trained with BA ensures better performance than GA.