Deep Learning 02 - Neural Networks

This talk discusses firstly shallow learning vs deep learning, i.e. linear vs non linear transformations. Second, the model of a neuron by Rosenblatt and its representation limits are presented. Then the Multi Layer Perceptron and its representation capabilities are presented and, afterwards, the backpropagation algorithm, i.e. the training algorithm of neural networks. Finally, the choice of a network size and a related simulation example are discussed.