Greedy layer-wise training of deep networks (2006)

by Yoshua Bengio, Pascal Lamblin, Dan Popovici, Hugo Larochelle