ModelGuard: Runtime Validation of Lipschitz-continuous Models

TL;DR

ModelGuard introduces a sampling-based runtime validation method for Lipschitz-continuous models, including black-box neural networks, providing probabilistic correctness guarantees and scalability demonstrated through case studies.

Contribution

It presents a novel sampling-based approach for runtime validation of Lipschitz-continuous models, applicable to black-box neural networks, with probabilistic correctness guarantees.

Findings

Effective validation of black-box Lipschitz models.

Provides confidence levels for trace correctness.

Scalable validation demonstrated in case studies.

Abstract

This paper presents ModelGuard, a sampling-based approach to runtime model validation for Lipschitz-continuous models. Although techniques exist for the validation of many classes of models the majority of these methods cannot be applied to the whole of Lipschitz-continuous models, which includes neural network models. Additionally, existing techniques generally consider only white-box models. By taking a sampling-based approach, we can address black-box models, represented only by an input-output relationship and a Lipschitz constant. We show that by randomly sampling from a parameter space and evaluating the model, it is possible to guarantee the correctness of traces labeled consistent and provide a confidence on the correctness of traces labeled inconsistent. We evaluate the applicability and scalability of ModelGuard in three case studies, including a physical platform.