I recently gave a presentation on the statistical mechanics of machine learning. This talk covers deep learning neural networks and how they are inspired by the physics of magnetism. Specifically, how the spin glass and Ising model are applied to study the hidden layers of deep neural networks.
Here is the presentation: