Neural Network Robustness Verification on GPUs

by   Christoph Muller, et al.

Certifying the robustness of neural networks against adversarial attacks is critical to their reliable adoption in real-world systems including autonomous driving and medical diagnosis. Unfortunately, state-of-the-art verifiers either do not scale to larger networks or are too imprecise to prove robustness, which limits their practical adoption. In this work, we introduce GPUPoly, a scalable verifier that can prove the robustness of significantly larger deep neural networks than possible with prior work. The key insight behind GPUPoly is the design of custom, sound polyhedra algorithms for neural network verification on a GPU. Our algorithms leverage the available GPU parallelism and the inherent sparsity of the underlying neural network verification task. GPUPoly scales to very large networks: for example, it can prove the robustness of a 1M neuron, 34-layer deep residual network in about 1 minute. We believe GPUPoly is a promising step towards the practical verification of large real-world networks.


Sound and Complete Verification of Polynomial Networks

Polynomial Networks (PNs) have demonstrated promising performance on fac...

Residual Error: a New Performance Measure for Adversarial Robustness

Despite the significant advances in deep learning over the past decade, ...

Simplifying Neural Networks with the Marabou Verification Engine

Deep neural network (DNN) verification is an emerging field, with divers...

Boosting Robustness Verification of Semantic Feature Neighborhoods

Deep neural networks have been shown to be vulnerable to adversarial att...

Neuro-Symbolic Verification of Deep Neural Networks

Formal verification has emerged as a powerful approach to ensure the saf...

Precise Multi-Neuron Abstractions for Neural Network Certification

Formal verification of neural networks is critical for their safe adopti...

Online Verification of Deep Neural Networks under Domain or Weight Shift

Although neural networks are widely used, it remains challenging to form...

Please sign up or login with your details

Forgot password? Click here to reset