mlsanityData to 14 Oct 2026

Investigating the Impact of Quantization on Adversarial Robustness

arXiv 2404.05639 ↗First posted 8 Apr 2024cs.LG cs.AI cs.CRcited by 9

Not on the map. Why?

Abstract (arXiv)

Quantization is a promising technique for reducing the bit-width of deep models to improve their runtime performance and storage efficiency, and thus becomes a fundamental step for deployment. In real-world scenarios, quantized models are often faced with adversarial attacks which cause the model to make incorrect inferences by introducing slight perturbations. However, recent studies have paid less attention to the impact of quantization on the model robustness. More surprisingly, existing studies on this topic even present inconsistent conclusions, which prompted our in-depth investigation. In this paper, we conduct a first-time analysis of the impact of the quantization pipeline components that can incorporate robust optimization under the settings of Post-Training Quantization and Quantization-Aware Training. Through our detailed analysis, we discovered that this inconsistency arises from the use of different pipelines in different studies, specifically regarding whether robust optimization is performed and at which quantization stage it occurs. Our research findings contribute insights into deploying more secure and robust quantized networks, assisting practitioners in reference for scenarios with high-security requirements and limited resources.

Key facts. Cited by 9 papers in the map (data through 14 Oct 2026). Builds on 25 of its 32 references that are in the map. First posted on arXiv 8 Apr 2024.

Builds on

25 of 32 references.

References · 25

  1. 2017Attention Is All You Needcited by 60,361
  2. 2015Deep Residual Learning for Image Recognitioncited by 52,096
  3. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,713
  4. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  5. 2017MobileNets: Efficient Convolutional Neural Networks for Mobile Vision Applicationscited by 3,845
  6. 2014Sequence to Sequence Learning with Neural Networkscited by 2,986
  7. 2020Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attackscited by 1,451
  8. 2018Obfuscated Gradients Give a False Sense of Security: Circumventing Defenses to Adversarial Examplescited by 1,268
  9. 2019Learned Step Size Quantizationcited by 628
  10. 2022ZeroQuant: Efficient and Affordable Post-Training Quantization for Large-Scale Transformerscited by 526
  11. 2021BRECQ: Pushing the Limit of Post-Training Quantization by Block Reconstructioncited by 479
  12. 2021A White Paper on Neural Network Quantizationcited by 472
13 more
  1. 2018PACT: Parameterized Clipping Activation for Quantized Neural Networkscited by 467
  2. 2019Defensive Quantization: When Efficiency Meets Robustnesscited by 95
  3. 2020Post-Training Piecewise Linear Quantization for Deep Neural Networkscited by 79
  4. 2022Mixed-Precision Neural Network Quantization via Learned Layer-wise Importancecited by 54
  5. 2017Attacking Binarized Neural Networkscited by 29
  6. 2020EMPIR: Ensembles of Mixed Precision Deep Networks for Increased Robustness against Adversarial Attackscited by 28
  7. 2023RobustMQ: Benchmarking Robustness of Quantized Modelscited by 19
  8. 2023A Survey on Approximate Edge AI for Energy Efficient Autonomous Driving Servicescited by 14
  9. 2022Understanding the Impact of Adversarial Robustness on Accuracy Disparitycited by 9
  10. 2023Benchmarking the Reliability of Post-training Quantization: a Particular Focus on Worst-case Performancecited by 9
  11. 2021Integer-arithmetic-only Certified Robustness for Quantized Neural Networkscited by 7
  12. 2022Transferring Adversarial Robustness Through Robust Representation Matchingcited by 7
  13. 2023SEAM: Searching Transferable Mixed-Precision Quantization Policy through Large Margin Regularizationcited by 6

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2025: 7; 2026: 2.

9 papers.

  1. 2026Can Large Language Models Still Explain Themselves? Investigating the Impact of Quantization on Self-Explanationscited by 5
  2. 2025How Quantization Shapes Bias in Large Language Modelscited by 4
  3. 2025Through a Compressed Lens: Investigating The Impact of Quantization on Factual Knowledge Recallcited by 3
  4. 2025Verification of Bit-Flip Attacks against Quantized Neural Networkscited by 2
  5. 2025Q-resafe: Assessing Safety Risks and Quantization-aware Safety Patching for Quantized Large Language Modelscited by 2
  6. 2025EAGER: Edge-Aligned LLM Defense for Robust, Efficient, and Accurate Cybersecurity Question Answeringcited by 2
  7. 2025Breaking the Limits of Quantization-Aware Defenses: QADT-R for Robustness Against Patch-Based Adversarial Attacks in QNNscited by 1
  8. 2025TriQDef: Disrupting Semantic and Gradient Alignment to Prevent Adversarial Patch Transferability in Quantized Neural Networkscited by 0
  9. 2026MOMAT: Mixture of Multiple Atlases for Low-Power Jailbreak Defense of Quantized LLMscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2023Benchmarking the Robustness of Quantized Models
  2. 2021A Layer-wise Adversarial-aware Quantization Optimization for Improving Robustness
  3. 2021On the Adversarial Robustness of Quantized Neural Networks
  4. 2021Qu-ANTI-zation: Exploiting Quantization Artifacts for Achieving Adversarial Outcomes
  5. 2023Improving the Robustness of Quantized Deep Neural Networks to White-Box Attacks using Stochastic Quantization and Information-Theoretic Ensemble Training
  6. 2022ODG-Q: Robust Quantization via Online Domain Generalization
  7. 2024Exploiting LLM Quantization
  8. 2025Smaller = Weaker? Benchmarking Robustness of Quantized LLMs in Code Generation