Multilayer Perceptron

To create more complex models, we can combine multiple perceptrons, all with their own set of weights and biases.

The outputs of a layer can be calculated with this formula, where the input values get a weight with $W$ , a bias value $b$ is added and the Sigmoid Activation Function will be applied to the resulting value.

y = \sigma(W \cdot x + b)

By combining multiple layers, using the outputs of former layers as new inputs, the following recursive formula applies:

y^{i+1} = \sigma(W^i \cdot y^i + b^i)