How to Train Multi-Layer Neural Nets

Jan Lecun says unsupervised learning is the missing link in getting us to higher forms of intelligent machines. He explains his version of self-supervised learning, called convolutional neural nets. The idea of multi-layer neural nets is that you can think of the first few layers as extracting features for the following layers to use. And if you can train the entire thing end to end, that means the system learns its own features.