MLP (Multilayer Perceptron)

  • Fully Connected layered architecture with feed-forward propagation of input signals.
  • Activation Functions are usually sigmoid and/or linear
  • Dynamics: simultaneous propagation of the signal with no delays, from the input layer to the output layer, traversing an arbitrary number of hidden layer (even 0 hidden layers is acceptable)
  • Learningsupervised, via the gradient method (backpropagation)

Original Files