An hour-long video summarizing the advances of ML in 2019, as well as some directions for 2020. Covers various topics from changes in ML frameworks to research breakthroughs.

NIPS 2019 workshop on combining DL and Bayesian models. Clarifies the conceptual differences, and introduces exciting ways to add Bayes to DL - e.g. uncertainty estimation in DL classifiers by using a more global view.

A very well-written and easy-to-understand post explaining the principle and purpose of Variational Autoencoders.

Fun paper on solving math problems - such as differential equations - with the powers of deep learning. This is actually *solving* the problems, not relying on things like gradient descent.