the presentation for the Spring School of Combinatorics 2016 on the paper "Mastering the Game of Go with Deep Neural Networks and Tree Search" by Google DeepMind
A bit off-topic, but highly interesting. Combining neural networks and Monte Carlo Tree Search, computers very recently defeated the first professional in the game of go. In short time, there is a scheduled match against the best human player. Good for someone with a background in machine learning/AI/etc. so that some overview and relevant details can be added. (The paper is to appear in Nature, the experiment is run by Google DeepMind.)