ICCV 2023 accepted papers on Adversarial robustness and security
51 of the 2,156 papers accepted at ICCV 2023 (2.4%) are on Adversarial robustness and security, compared with 3.5% at ICCV 2021. They fall into 4 topics.
See ICCV 2023 in the interactive map
Submitting to ICCV? Checklist →
Working on something in this area? Describe your idea in scime atlas to see the closest papers by meaning and how crowded the spot has become.
Topics and papers
Attacks on text and graphs · 19 papers
- Set-level Guidance Attack: Boosting Adversarial Transferability of Vision-Language Pre-training Models (103 citations)
- AdvDiffuser: Natural Adversarial Example Synthesis with Diffusion Models (80 citations)
- Robust Evaluation of Diffusion-Based Adversarial Purification (70 citations)
- ACTIVE: Towards Highly Transferable 3D Physical Camouflage for Universal and Robust Vehicle Evasion (47 citations)
- Unified Adversarial Patch for Cross-Modal Attacks in the Physical World (47 citations)
- Among Us: Adversarially Robust Collaborative Perception by Consensus (45 citations)
- RFLA: A Stealthy Reflected Light Adversarial Attack in the Physical World (44 citations)
- Does Physical Adversarial Example Really Matter to Autonomous Driving? Towards System-Level Effect of Adversarial Object Evasion Attack (43 citations)
- Practical Membership Inference Attacks Against Large-Scale Multi-Modal Models: A Pilot Study (34 citations)
- Downstream-agnostic Adversarial Examples (31 citations)
- REAP: A Large-Scale Realistic Adversarial Patch Benchmark (31 citations)
- Towards Viewpoint-Invariant Visual Recognition via Adversarial Training (25 citations)
and 7 more in the interactive map.
Certified and empirical robustness · 16 papers
- Improving Generalization of Adversarial Training via Robust Critical Fine-Tuning (45 citations)
- Robust Mixture-of-Expert Training for Convolutional Neural Networks (40 citations)
- SAFARI: Versatile and Efficient Evaluations for Robustness of Interpretability (40 citations)
- Enhancing Generalization of Universal Adversarial Perturbation through Gradient Aggregation (36 citations)
- Semi-supervised Semantics-guided Adversarial Training for Robust Trajectory Prediction (28 citations)
- HybridAugment++: Unified Frequency Spectra Perturbations for Model Robustness (21 citations)
- Towards Building More Robust Models with Frequency Bias (20 citations)
- Fast Adversarial Training with Smooth Convergence (16 citations)
- MUter: Machine Unlearning on Adversarially Trained Models (16 citations)
- LEA2: A Lightweight Ensemble Adversarial Attack via Non-overlapping Vulnerable Frequency Regions (14 citations)
- Improving Adversarial Robustness of Masked Autoencoders via Test-time Frequency-domain Prompting (13 citations)
- Advancing Example Exploitation Can Alleviate Critical Challenges in Adversarial Training (9 citations)
and 4 more in the interactive map.
Black-box and transfer attacks · 11 papers
- Structure Invariant Transformation for better Adversarial Transferability (118 citations)
- An Adaptive Model Ensemble Adversarial Attack for Boosting Adversarial Transferability (99 citations)
- Boosting Adversarial Transferability via Gradient Relevance Attack (77 citations)
- Transferable Adversarial Attack for Both Vision Transformers and Convolutional Networks via Momentum Integrated Gradients (51 citations)
- CGBA: Curvature-aware Geometric Black-box Attack (32 citations)
- Efficient Decision-based Black-box Patch Attacks on Video Recognition (32 citations)
- TRM-UAP: Enhancing the Transferability of Data-Free Universal Adversarial Perturbation via Truncated Ratio Maximization (16 citations)
- Hard No-Box Adversarial Attack on Skeleton-Based Human Action Recognition with Skeleton-Motion-Informed Gradient (15 citations)
- Backpropagation Path Search On Adversarial Transferability (6 citations)
- How to Choose your Best Allies for a Transferable Attack? (2 citations)
- Tracing the Origin of Adversarial Attack for Forensic Investigation and Deterrence (2 citations)
Backdoor and poisoning · 5 papers
- CleanCLIP: Mitigating Data Poisoning Attacks in Multimodal Contrastive Learning (61 citations)
- One-bit Flip is All You Need: When Bit-flip Attack Meets Model Training (26 citations)
- Computation and Data Efficient Backdoor Attacks (15 citations)
- Beating Backdoor Attack at Its Own Game (12 citations)
- The Victim and The Beneficiary: Exploiting a Poisoned Model to Train a Clean Model on Poisoned Data (3 citations)
