The ability to learn and generalize is fundamental to any learning machine. In particular, generalization behavior is one of the most important topics in any classifier trained non-parametrically (e.g., neural networks). Artificial neural networks (ANNs) use inductive learning to find general concepts from their concrete examples. If there is set of input output to pairs called training set, the network parameters is optimized in order to fit network output to the given input. The fit is evaluated by means of cost function, usually assumed to be the mean square error. On-line training algorithms adapt the network parameters to the changing data statistics. After training, the network can be used with data whose underlying statistics is similar to the training set.
In this MATLAB code, we use genetic algorithm for training MLP.
Genetic algorithm find optimum weight of MLP.