MATLAB Simulation of Perceptron Learning

In machine learning, the perceptron is an algorithm for supervised learning of binary classifiers (functions that can decide whether an input, represented by a vector of numbers, belongs to some specific class or not).It is a type of linear classifier, i.e. a classification algorithm that makes its predictions based on a linear predictor function combining a set of weights with the feature vector.

In short the hidden layer provides non-linearity. You can think of each hidden neuron as a single logistic regression. Each logistic regression has a linear decision boundary. With more than one linear, non-parallel lines, you can draw a convex boundary - more lines, more flexibility. The combined boundary is the final output layer.