CVPR 2023 accepted papers on Adversarial robustness and security
75 of the 2,359 papers accepted at CVPR 2023 (3.2%) are on Adversarial robustness and security, compared with 3.0% at CVPR 2022. They fall into 4 topics.
See CVPR 2023 in the interactive map
Working on something in this area? Describe your idea in scime atlas to see the closest accepted papers by meaning and how crowded the spot has become.
Topics and papers
Attacks on text and graphs · 27 papers
- Evading DeepFake Detectors via Adversarial Statistical Consistency (67 citations)
- TrojDiff: Trojan Attacks on Diffusion Models With Diverse Targets (60 citations)
- Re-Thinking Model Inversion Attacks Against Deep Neural Networks (58 citations)
- Physically Adversarial Infrared Patches With Learnable Shapes and Locations (49 citations)
- Jedi: Entropy-Based Localization and Removal of Adversarial Patches (47 citations)
- Robust Single Image Reflection Removal Against Adversarial Attacks (40 citations)
- Privacy-Preserving Adversarial Facial Features (39 citations)
- Physical-World Optical Adversarial Attacks on 3D Face Recognition (36 citations)
- Adversarial Counterfactual Visual Explanations (34 citations)
- SlowLiDAR: Increasing the Latency of LiDAR-Based Detection Using Adversarial Examples (33 citations)
- ImageNet-E: Benchmarking Neural Network Robustness via Attribute Editing (26 citations)
- Can't Steal? Cont-Steal! Contrastive Stealing Attacks Against Image Encoders (24 citations)
and 15 more in the interactive map.
Certified and empirical robustness · 21 papers
- Boosting Accuracy and Robustness of Student Models via Adaptive Adversarial Distillation (55 citations)
- CFA: Class-Wise Calibrated Fair Adversarial Training (44 citations)
- Randomized Adversarial Training via Taylor Expansion (44 citations)
- Revisiting Residual Networks for Adversarial Robustness (38 citations)
- The Enemy of My Enemy Is My Friend: Exploring Inverse Adversaries for Improving Adversarial Training (37 citations)
- Feature Separation and Recalibration for Adversarial Robustness (36 citations)
- Towards Compositional Adversarial Robustness: Generalizing Adversarial Training to Composite Semantic Perturbations (34 citations)
- The Best Defense Is a Good Offense: Adversarial Augmentation Against Adversarial Attacks (26 citations)
- BiasAdv: Bias-Adversarial Augmentation for Model Debiasing (25 citations)
- Trade-Off Between Robustness and Accuracy of Vision Transformers (22 citations)
- CUDA: Convolution-Based Unlearnable Datasets (20 citations)
- Demystifying Causal Features on Adversarial Examples and Causal Inoculation for Robust Network by Adversarial Instrumental Variable Regression (20 citations)
and 9 more in the interactive map.
Backdoor and poisoning · 14 papers
- Color Backdoor: A Robust Poisoning Attack in Color Space (87 citations)
- How to Backdoor Diffusion Models? (77 citations)
- Backdoor Defense via Adaptively Splitting Poisoned Dataset (60 citations)
- TrojViT: Trojan Insertion in Vision Transformers (46 citations)
- Backdoor Defense via Deconfounded Representation Learning (43 citations)
- Backdoor Attacks Against Deep Image Compression via Adaptive Frequency Trigger (42 citations)
- You Are Catching My Attention: Are Vision Transformers Bad Learners Under Backdoor Attacks? (36 citations)
- Detecting Backdoors During the Inference Stage Based on Corruption Robustness Consistency (34 citations)
- Backdoor Cleansing With Unlabeled Data (24 citations)
- Defending Against Patch-Based Backdoor Attacks on Self-Supervised Learning (24 citations)
- Architectural Backdoors in Neural Networks (18 citations)
- Progressive Backdoor Erasing via Connecting Backdoor and Adversarial Attacks (18 citations)
and 2 more in the interactive map.
Black-box and transfer attacks · 13 papers
- Transferable Adversarial Attacks on Vision Transformers With Token Gradient Regularization (82 citations)
- Improving the Transferability of Adversarial Samples by Path-Augmented Method (61 citations)
- Towards Transferable Targeted Adversarial Examples (54 citations)
- Sibling-Attack: Rethinking Transferable Adversarial Attacks Against Face Recognition (53 citations)
- Towards Effective Adversarial Textured 3D Meshes on Physical Face Recognition (48 citations)
- Enhancing the Self-Universality for Transferable Targeted Attacks (44 citations)
- Reinforcement Learning-Based Black-Box Model Inversion Attacks (36 citations)
- StyLess: Boosting the Transferability of Adversarial Examples (34 citations)
- RIATIG: Reliable and Imperceptible Adversarial Text-to-Image Generation With Natural Prompts (32 citations)
- Introducing Competition To Boost the Transferability of Targeted Adversarial Examples Through Clean Feature Mixup (29 citations)
- Minimizing Maximum Model Discrepancy for Transferable Black-Box Targeted Attacks (16 citations)
- Boosting Verified Training for Robust Image Classifications via Abstraction (10 citations)
and 1 more in the interactive map.