We consider the task of learning a classifier from the feature space $\mathcal{ X}$ to the set of classes $\mathcal{ Y} = \{ 0, 1\}$, when the features can be partitioned into class-conditionally independent feature sets $\mathcal{ X}_1$ and $\mathcal{ X}_2$. We show the surprising fact that the class-conditional independence can be used to represent the original learning task in terms of 1) learning a classifier from $\mathcal{ X}_2$ to $\mathcal{ X}_1$ and 2) learning the class-conditional distribution of the feature set $\mathcal{ X}_1$. This fact can be exploited for semi-supervised learning because the former task can be accomplished purely from unlabeled samples. We present experimental evaluation of the idea in two real world applications.
Categories: cs.LG