Artificial Neural Networks – Part 2: MLP Implementation for XOr

As promised in part one, this second part details a java implementation of a multilayer perceptron (MLP) for the XOr problem. Actually, as you will see, the core classes are designed to implement any MLP implementation with a single hidden layer. First, it will help to introduce a quick overview of how MLP networks can be used to make predictions for the XOr problem. For a more detailed explanation, please review part one of this post. The image at the top of this article depicts the architecture for a multilayer perceptron network designed specifically to solve the XOr problem.