Research map / Adversarial robustness and security / Backdoor and poisoning
Targeted Attack against Deep Neural Networks via Flipping Limited Weight Bits
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Optimize binary weight-bit flips with a cardinality constraint instead of using heuristic bit selection | targeted attacks on quantized DNNs · CIFAR-10 and ImageNet | N_flip: 5.57 {$\pm$1.58} vs 9.91 {$\pm$2.33} on CIFAR-10 ResNet 8-bit; 7.37 {$\pm$2.18} vs 18.54 {$\pm$6.14} on ImageNet ResNet | Helped |
| Use piece-wise clustering defense | TA-LBF on CIFAR-10 ResNet 8-bit · CIFAR-10 | N_flip: 5.57 {$\pm$1.58} -> 18.93 {$\pm$7.11} | Hurt |
| Increase network width to 2x | TA-LBF on CIFAR-10 ResNet 8-bit · CIFAR-10 | N_flip: 5.57 {$\pm$1.58} -> 8.79 {$\pm$2.44} | Hurt |
| Increase lambda while holding k fixed (ablation) | TA-LBF · CIFAR-10, ResNet 8-bit | PA-ACC | Helped |
| Increase lambda while holding k fixed (ablation) | TA-LBF · CIFAR-10, ResNet 8-bit | ASR | Hurt |
| Increase the bit-flip budget k (ablation) | TA-LBF · CIFAR-10, ResNet 8-bit | N_flip | Mixed |
| Use more auxiliary samples, particularly increasing N from below 200 (ablation) | TA-LBF · CIFAR-10, ResNet 8-bit | PA-ACC | Helped |
Closest papers by meaning
- Crossfire: An Elastic Defense Framework for Graph Neural Networks Under Bit Flip Attacks (AAAI 2025 · similarity 0.85)
- Train to Defend: First Defense Against Cryptanalytic Neural Network Parameter Extraction Attacks (NeurIPS 2025 · similarity 0.85)
- Your Scale Factors are My Weapon: Targeted Bit-Flip Attacks on Vision Transformers via Scale Factor Manipulation (CVPR 2025 · similarity 0.84)
- Trap-MID: Trapdoor-based Defense against Model Inversion Attacks (NeurIPS 2024 · similarity 0.84)
- Deep-TROJ: An Inference Stage Trojan Insertion Algorithm through Efficient Weight Replacement Attack (CVPR 2024 · similarity 0.83)
- Flexible, Efficient, and Stable Adversarial Attacks on Machine Unlearning (ICML 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.