Section outline

  • Introduction to deep neural networks: non-linearly separable problems, training from data and representation learning. The multi-level perceptron and the Back-propagation algorithm. Convolutional and recurrent neural networks. The very deep approaches.