Towards Efficient Verification of Quantized Neural Networks

Abstract

Quantization replaces floating point arithmetic with integer arithmetic in deep neural network models, providing more efficient on-device inference with less power and memory. In this work, we propose a framework for formally verifying the properties of quantized neural networks. Our baseline technique is based on integer linear programming which guarantees both soundness and completeness. We then show how efficiency can be improved by utilizing gradient-based heuristic search methods and also bound-propagation techniques. We evaluate our approach on perception networks quantized with PyTorch. Our results show that we can verify quantized networks with better scalability and efficiency than the previous state of the art.

Cite

Text

Huang et al. "Towards Efficient Verification of Quantized Neural Networks." AAAI Conference on Artificial Intelligence, 2024. doi:10.1609/AAAI.V38I19.30108

Markdown

[Huang et al. "Towards Efficient Verification of Quantized Neural Networks." AAAI Conference on Artificial Intelligence, 2024.](https://mlanthology.org/aaai/2024/huang2024aaai-efficient/) doi:10.1609/AAAI.V38I19.30108

BibTeX

@inproceedings{huang2024aaai-efficient,
  title     = {{Towards Efficient Verification of Quantized Neural Networks}},
  author    = {Huang, Pei and Wu, Haoze and Yang, Yuting and Daukantas, Ieva and Wu, Min and Zhang, Yedi and Barrett, Clark W.},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {2024},
  pages     = {21152-21160},
  doi       = {10.1609/AAAI.V38I19.30108},
  url       = {https://mlanthology.org/aaai/2024/huang2024aaai-efficient/}
}