● mastered · kind model · level 4 · 30h

Stacked fully-connected layers with non-linear activations — a universal function approximator. The baseline every specialised architecture refines.

In deep learning, a multilayer Perceptron (MLP) is a kind of modern feedforward Neural Network consisting of fully connected neurons with nonlinear Activation Functions, organized in layers, notable for being able to distinguish data that is not linearly separable.

Enlaces

Fuentes