Incremental Neural Network Verification via Learned Conflicts

Elsaleh, Davis, Wu & Katz (2026). A technically elegant paper that applies incremental SAT-style conflict reuse to neural network verification. Rather than solving each verification query from scratch, the verifier caches learned infeasible activation phase combinations and inherits them across related queries, yielding speedups of up to 1.9x. Directly applicable to safety-critical AI deployment.