Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

PRIMA: General and Precise Neural Network Certification via Scalable Convex Hull Approximations

About

Formal verification of neural networks is critical for their safe adoption in real-world applications. However, designing a precise and scalable verifier which can handle different activation functions, realistic network architectures and relevant specifications remains an open and difficult challenge. In this paper, we take a major step forward in addressing this challenge and present a new verification framework, called PRIMA. PRIMA is both (i) general: it handles any non-linear activation function, and (ii) precise: it computes precise convex abstractions involving multiple neurons via novel convex hull approximation algorithms that leverage concepts from computational geometry. The algorithms have polynomial complexity, yield fewer constraints, and minimize precision loss. We evaluate the effectiveness of PRIMA on a variety of challenging tasks from prior work. Our results show that PRIMA is significantly more precise than the state-of-the-art, verifying robustness to input perturbations for up to 20%, 30%, and 34% more images than existing work on ReLU-, Sigmoid-, and Tanh-based networks, respectively. Further, PRIMA enables, for the first time, the precise verification of a realistic neural network for autonomous driving within a few minutes.

Mark Niklas M\"uller, Gleb Makarchuk, Gagandeep Singh, Markus P\"uschel, Martin Vechev• 2021

Related benchmarks

TaskDatasetResultRank
Neural Network VerificationMNIST 1000 images
Verified Accuracy77.5
30
Robustness VerificationCIFAR-10 epsilon=2/255 (first 200 examples)
Verified Accuracy53.5
20
Robust VerificationMNIST (test)
Average Time (s)135.9
17
Neural Network VerificationCIFAR 1000 images
Verified Accuracy48.3
13
Neural Network Verificationoval VNN-COMP 2022
Verification Time (s)233.8
10
Robustness VerificationMNIST epsilon=0.3 (first 200 examples)
Verified Accuracy44.5
8
Showing 6 of 6 rows

Other info

Follow for update