Scalable Verification of Quantized Neural Networks

Authors

  • Thomas A. Henzinger IST Austria
  • Mathias Lechner IST Austria
  • Đorđe Žikelić IST Austria

Keywords:

Satisfiability Modulo Theories

Abstract

Formal verification of neural networks is an active topic of research, and recent advances have significantly increased the size of the networks that verification tools can handle. However, most methods are designed for verification of an idealized model of the actual network which works over real arithmetic and ignores rounding imprecisions. This idealization is in stark contrast to network quantization, which is a technique that trades numerical precision for computational efficiency and is, therefore, often applied in practice. Neglecting rounding errors of such low-bit quantized neural networks has been shown to lead to wrong conclusions about the network's correctness. Thus, the desired approach for verifying quantized neural networks would be one that takes these rounding errors into account. In this paper, we show that verifying the bit-exact implementation of quantized neural networks with bit-vector specifications is PSPACE-hard, even though verifying idealized real-valued networks and satisfiability of bit-vector specifications alone are each in NP. Furthermore, we explore several practical heuristics toward closing the complexity gap between idealized and bit-exact verification. In particular, we propose three techniques for making SMT-based verification of quantized neural networks more scalable. Our experiments demonstrate that our proposed methods allow a speedup of up to three orders of magnitude over existing approaches.

Downloads

Published

2021-05-18

How to Cite

Henzinger, T. A., Lechner, M., & Žikelić, Đorđe. (2021). Scalable Verification of Quantized Neural Networks. Proceedings of the AAAI Conference on Artificial Intelligence, 35(5), 3787-3795. Retrieved from https://ojs.aaai.org/index.php/AAAI/article/view/16496

Issue

Section

AAAI Technical Track on Constraint Satisfaction and Optimization