Talking with Verifiers: Automatic Specification Generation for Neural Network Verification

TL;DR

This paper introduces a framework that translates natural language specifications into formal verification queries, enabling neural network verification tools to handle high-level, semantic correctness requirements more effectively.

Contribution

It presents a novel method for automatically converting natural language specifications into formal verification queries, broadening the scope of neural network verification tools.

Findings

Successfully verifies complex semantic specifications

Maintains high fidelity to user intent

Low computational overhead

Abstract

Neural network verification tools currently support only a narrow class of specifications, typically expressed as low-level constraints over raw inputs and outputs. This limitation significantly hinders their adoption and practical applicability across diverse application domains where correctness requirements are naturally expressed at a higher semantic level. This challenge is rooted in the inherent nature of deep neural networks, which learn internal representations that lack an explicit mapping to human-understandable features. To address this, we bridge this gap by introducing a novel component to the verification pipeline, making existing verification tools applicable to a broader range of domains and specification styles. Our framework enables users to formulate specifications in natural language, which are then automatically analyzed and translated into formal verification queries compatible with state-of-the-art neural network verifiers. We evaluate our approach on both structured and unstructured datasets, demonstrating that it successfully verifies complex semantic specifications that were previously inaccessible. Our results show that this translation process maintains high fidelity to user intent while incurring low computational overhead, thereby substantially extending the applicability of formal DNN verification to real-world, high-level requirements.