Neural Networks and Deep Learning
How does the brain do it?
If features are complex enough, anything can be classified?
Single neurons are not able to solve complex tasks (linear decision boundaries). More layers of linear units are not enough (still linear).
How can we learn the weights? Theoretical result [Cybenko, 1989]: 2-layer net with linear output can approximate any continuous function over compact domain to arbitrary accuracy (given enough hidden units!)
Why use Deep Multi Layered Models? Is there a theoretical justification? No
Neural Network examples Standard NN Convolutional NN Recurrent NN
No closed form solution for the Maximum Likelihood for this model!
Feed forward Networks
Books and Resources We will mostly follow Deep Learning by Ian Goodfellow,Yoshua Bengio and Aaron Courville (MIT Press, 2016) Stanford CS 231n: by Li, Karpathy & Johnson Neural Networks and Deep Learning by Michael Nielsen Bishop - Pattern Recognition And Machine Learning - Springer 2006 Uncertainty in Deep Learning Yarin Gal Department of Engineering University of Cambridge
Books and Resources Probabilistic machine learning and artificial intelligence Zoubin Ghahramani1