Factorial Learning by Clustering Features
Joshua B. Tenenbaum, Emanuel Todorov · 1994
We introduce a novel algorithm for factorial learning, motivated by segmentation problems in computational vision, in which the underlying factors correspond to clusters of highly correlated input features. The algorithm derives from a new kind of competitive clustering model, in which the cluster generators compete to ex-plain each feature of the data set and cooperate to explain each input example, rather than competing for examples and cooper-ating on features, as in traditional clustering algorithms. A natu-ral extension of the algorithm recovers hierarchical models of data generated from multiple unknown categories, each with a differ-ent, multiple causal structure. Several simulations demonstrate the power of this approach. 1