23
0

Learnability of Parameter-Bounded Bayes Nets

Abstract

Bayes nets are extensively used in practice to efficiently represent joint probability distributions over a set of random variables and capture dependency relations. In a seminal paper, Chickering et al. (JMLR 2004) showed that given a distribution P\mathbb{P}, that is defined as the marginal distribution of a Bayes net, it is NP\mathsf{NP}-hard to decide whether there is a parameter-bounded Bayes net that represents P\mathbb{P}. They called this problem LEARN. In this work, we extend the NP\mathsf{NP}-hardness result of LEARN and prove the NP\mathsf{NP}-hardness of a promise search variant of LEARN, whereby the Bayes net in question is guaranteed to exist and one is asked to find such a Bayes net. We complement our hardness result with a positive result about the sample complexity that is sufficient to recover a parameter-bounded Bayes net that is close (in TV distance) to a given distribution P\mathbb{P}, that is represented by some parameter-bounded Bayes net, generalizing a degree-bounded sample complexity result of Brustle et al. (EC 2020).

View on arXiv
Comments on this paper