Quantization Blindspots: How Model Compression Breaks Backdoor Defenses
By: Rohan Pandey, Eric Ye
Potential Business Impact:
Makes AI security tricks stop working on phones.
Backdoor attacks embed input-dependent malicious behavior into neural networks while preserving high clean accuracy, making them a persistent threat for deployed ML systems. At the same time, real-world deployments almost never serve full-precision models: post-training quantization to INT8 or lower precision is now standard practice for reducing memory and latency. This work asks a simple question: how do existing backdoor defenses behave under standard quantization pipelines? We conduct a systematic empirical study of five representative defenses across three precision settings (FP32, INT8 dynamic, INT4 simulated) and two standard vision benchmarks using a canonical BadNet attack. We observe that INT8 quantization reduces the detection rate of all evaluated defenses to 0% while leaving attack success rates above 99%. For INT4, we find a pronounced dataset dependence: Neural Cleanse remains effective on GTSRB but fails on CIFAR-10, even though backdoors continue to survive quantization with attack success rates above 90%. Our results expose a mismatch between how defenses are commonly evaluated (on FP32 models) and how models are actually deployed (in quantized form), and they highlight quantization robustness as a necessary axis in future evaluations and designs of backdoor defenses.
Similar Papers
A Quantized VAE-MLP Botnet Detection Model: A Systematic Evaluation of Quantization-Aware Training and Post-Training Quantization Strategies
Machine Learning (CS)
Makes small devices catch internet hackers faster.
Quantitative Analysis of Deeply Quantized Tiny Neural Networks Robust to Adversarial Attacks
Machine Learning (CS)
Makes smart programs smaller and safer from tricks.
InfiJanice: Joint Analysis and In-situ Correction Engine for Quantization-Induced Math Degradation in Large Language Models
Machine Learning (CS)
Fixes AI math mistakes after shrinking it.