Abstraction-Based Proof Production in Formal Verification of Neural Networks
This work addresses the problem of increasing reliability in formal verification of neural networks for developers and researchers, though it is preliminary and incremental in integrating abstraction with proof production.
The paper tackles the gap between scalability and provable guarantees in neural network verification by introducing a framework for proof-producing abstraction-based verification, which modularly separates the task into proving abstract network correctness and abstraction soundness, with the latter being a novel method for generating formal proofs.
Modern verification tools for deep neural networks (DNNs) increasingly rely on abstraction to scale to realistic architectures. In parallel, proof production is becoming a critical requirement for increasing the reliability of DNN verification results. However, current proofproducing verifiers do not support abstraction-based reasoning, creating a gap between scalability and provable guarantees. We address this gap by introducing a novel framework for proof-producing abstraction-based DNN verification. Our approach modularly separates the verification task into two components: (i) proving the correctness of an abstract network, and (ii) proving the soundness of the abstraction with respect to the original DNN. The former can be handled by existing proof-producing verifiers, whereas we propose the first method for generating formal proofs for the latter. This preliminary work aims to enable scalable and trustworthy verification by supporting common abstraction techniques within a formal proof framework.