Compositional Noisy-Logical Learning

We describe a new method for learning the conditional probability distribution of a binary-valued variable from labelled training examples. Our proposed Compositional Noisy-Logical Learning (CNLL) a