Learning CNF formulas from uniform random solutions in the local lemma regime
Data Structures and Algorithms
2025-11-05 v1 Machine Learning
Machine Learning
Abstract
We study the problem of learning a -variables -CNF formula from its i.i.d. uniform random solutions, which is equivalent to learning a Boolean Markov random field (MRF) with -wise hard constraints. Revisiting Valiant's algorithm (Commun. ACM'84), we show that it can exactly learn (1) -CNFs with bounded clause intersection size under Lov\'asz local lemma type conditions, from samples; and (2) random -CNFs near the satisfiability threshold, from samples. These results significantly improve the previous sample complexity. We further establish new information-theoretic lower bounds on sample complexity for both exact and approximate learning from i.i.d. uniform random solutions.
Cite
@article{arxiv.2511.02487,
title = {Learning CNF formulas from uniform random solutions in the local lemma regime},
author = {Weiming Feng and Xiongxin Yang and Yixiao Yu and Yiyao Zhang},
journal= {arXiv preprint arXiv:2511.02487},
year = {2025}
}