Patch-based adversarial attacks remain highly effective and transferable across quantized neural networks, and a proposed quantization-aware defense training method substantially reduces their success rate.
Double-win quant: Aggressively winning ro- bustness of quantized deep neural networks via random pre- cision training and inference
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CR 1years
2024 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Exploring the Robustness and Transferability of Patch-Based Adversarial Attacks in Quantized Neural Networks
Patch-based adversarial attacks remain highly effective and transferable across quantized neural networks, and a proposed quantization-aware defense training method substantially reduces their success rate.