Spring workshop on Mining and Learning (SML) edition:1 location:Traben-Trarbach date:21-25 April 2008
Causal Probabilistic logic (CP-logic) is a language for describing complex probabilistic processes. In this talk we consider the problem of learning CP-theories from data. We briefly discuss three possible approaches. First, we review the existing algorithm by Meert et al. Second, we show how simple CP-theories can be learned by using the learning algorithm for Logical Bayesian Networks and converting the result into a CP-theory. Third, we argue that for learning more complex CP-theories, an algorithm that combines the ideas behind the two previous approaches might work best.