Perceptron is a basic linear classifier that outputs binary labels.
Perceptron is a basic linear classifier that outputs binary labels. If the training data set is not linear separable, the learning algorithm cannot converge.
A classical problem of XOR is a dataset that is not linear separable. A perceptron does not work in this case. By adding nonlinear layers between the input and output, one can separate all data. With enough training data, the resulting network is able to model any well-defined function to arbitrary precision. This model is a generalization known as a multilayer perceptron.
For more details about perceptron, see wiki.
