A new branch of software is emerging from artificial intelligence research, distinguished from machine learning.
Give this a listen, its good stuff:
The difference between deep learning and machine learning is huge. To over-simplify things, machine learning is about “training” software by showing it many hundreds or thousands of examples of something, then guiding and correcting the conclusions the software makes based on prior learning. It’s very manually intensive.
Deep learning, on the other hand, is about letting the software train itself and only correcting it where necessary. In the interview the researcher talked about feeding one of his applications – perhaps the truest meaning of application – about 1.5 million random pictures. After that, it can tell the difference not just between a dog and a cat, but even it’s breed, even if it’s under furniture. Compared to what machine learning can do, that’s out-of-the-park sophisticated.