We're mining a rich vein of techniques that have been opened up by the deep learning revolution. There doesn't seem to be any near-term end in sight, and the pa

We're mining a rich vein of techniques that have been opened up by the deep learning revolution. There doesn't seem to be any near-term end in sight, and the pace of improvement is incredible. It's hard just to keep up with the best result in, say, computer vision or NLP. This is nothing like the 80's where techniques foundered on non-toy problems.