Learning Lookahead Lemmas for Neural Network Verification

Researchers Liam Davis and Haoze Wu introduced an inprocessing framework for neural network verification that uses lookahead to derive lemmas over unstable ReLUs, improving the performance of state-of-the-art verifiers Marabou and α-β-CROWN by proving up to 34% more instances unsatisfiable.

Learning Lookahead Lemmas for Neural Network Verification By Liam Davis, Haoze WuSource: arXiv cs.LG https://arxiv.org/list/cs.LG/recent arXiv:2607.29051v1 Announce Type: new Abstract: State-of-the-art neural network /glossary/neural-network verifiers use the branch-and-bound procedure as their core solving mechanism. We introduce an inprocessing framework for neural network verification driven by the lookahead procedure. Under this framework, lookahead derives new lemmas over the phases of unstable ReLUs, which are collected into an implication graph that is used to prune the search space and vivify boolean cuts. We instantiate the framework in two state-of-the-art verifiers, Marabou and $\alpha$-$\beta$-CROWN, and demonstrate that it improves performance in both, proving up to 34% more instances unsatisfiable.Get AI news in your inbox Daily digest of what matters in AI.