A fast learning algorithm for deep belief nets
At a glance
Made deep networks easier to train by teaching them one layer at a time.
In one paragraph
Made training deep neural networks more practical by teaching them one layer at a time, then refining the whole network to recognize handwritten digits.
