Unsupervised Learning by Discriminating Data from Artificial Noise
Published on Mar 26, 20104187 Views
Noise-contrastive estimation is a new estimation principle that we have developed for parameterized statistical models. The idea is to train a classifier to discriminate between the observed data and