An artificial neural network is an advanced classifier inspired by the human brain. It is an ensemble of simple classifiers
working together. (Click here to play with simple classifiers).
By combining multiple linear decision boundaries the ensemble has the ability to model any shape decision boundary.
In the example above, the inputs x1 and x2 are fed into two hidden neurons (classifiers). The first divides the feature space with a vertical decision
boundary, the second with a hortizonal boundary. These are fed into an output neuron which combines their decisions creating the
nonlinear decision boundary pictured.
Click here to play with an application of neural networks.