NeurIPS 2022 accepted papers on Adversarial robustness and security
105 of the 2,901 papers accepted at NeurIPS 2022 (3.6%) are on Adversarial robustness and security, compared with 3.6% at NeurIPS 2021. They fall into 4 topics.
See NeurIPS 2022 in the interactive map
Submitting to NeurIPS? 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
Certified and empirical robustness · 57 papers
- Make Some Noise: Reliable and Efficient Single-Step Adversarial Training (31 citations)
- SNN-RAT: Robustness-enhanced Spiking Neural Network through Regularized Adversarial Training (28 citations)
- When Adversarial Training Meets Vision Transformers: Recipes from Training to Architecture (14 citations)
- Rethinking Lipschitz Neural Networks and Certified Robustness: A Boolean Function Perspective (12 citations)
- Efficient and Effective Augmentation Strategy for Adversarial Training (11 citations)
- DISCO: Adversarial Defense with Local Implicit Functions (10 citations)
- Robust Feature-Level Adversaries are Interpretability Tools (9 citations)
- A2: Efficient Automated Attacker for Boosting Adversarial Training (8 citations)
- Toward Robust Spiking Neural Network Against Adversarial Perturbation (8 citations)
- Label Noise in Adversarial Training: A Novel Perspective to Study Robust Overfitting (6 citations)
- Explicit Tradeoffs between Adversarial and Natural Distributional Robustness (5 citations)
- Exploiting the Relationship Between Kendall's Rank Correlation and Cosine Similarity for Attribution Protection (5 citations)
and 45 more in the interactive map.
Attacks on text and graphs · 24 papers
- Adv-Attribute: Inconspicuous and Transferable Adversarial Attack on Face Recognition (23 citations)
- Perceptual Attacks of No-Reference Image Quality Models with Human-in-the-Loop (18 citations)
- Isometric 3D Adversarial Examples in the Physical World (17 citations)
- Practical Adversarial Attacks on Spatiotemporal Traffic Forecasting Models (15 citations)
- Revisiting Injective Attacks on Recommender Systems (12 citations)
- Are Defenses for Graph Neural Networks Robust? (11 citations)
- Are You Stealing My Model? Sample Correlation for Fingerprinting Deep Neural Networks (10 citations)
- Enhance the Visual Representation via Discrete Adversarial Training (10 citations)
- SALSA: Attacking Lattice Cryptography with Transformers (10 citations)
- Towards Reasonable Budget Allocation in Untargeted Graph Structure Attacks via Gradient Debias (10 citations)
- Adversarial training for high-stakes reliability (9 citations)
- GAMA: Generative Adversarial Multi-Object Scene Attacks (9 citations)
and 12 more in the interactive map.
Backdoor and poisoning · 17 papers
- BackdoorBench: A Comprehensive Benchmark of Backdoor Learning (41 citations)
- Untargeted Backdoor Watermark: Towards Harmless and Stealthy Dataset Copyright Protection (25 citations)
- A Unified Evaluation of Textual Backdoor Learning: Frameworks and Benchmarks (24 citations)
- Handcrafted Backdoors in Deep Neural Networks (19 citations)
- Rethinking the Reverse-engineering of Trojan Triggers (19 citations)
- BadPrompt: Backdoor Attacks on Continuous Prompts (18 citations)
- Training with More Confidence: Mitigating Injected and Natural Backdoors During Training (12 citations)
- Effective Backdoor Defense by Exploiting Sensitivity of Poisoned Samples (11 citations)
- Marksman Backdoor: Backdoor Attacks with Arbitrary Target Class (10 citations)
- Sleeper Agent: Scalable Hidden Trigger Backdoors for Neural Networks Trained from Scratch (10 citations)
- Pre-activation Distributions Expose Backdoor Neurons (4 citations)
- One-shot Neural Backdoor Erasing via Adversarial Weight Masking (3 citations)
and 5 more in the interactive map.
Black-box and transfer attacks · 7 papers
- Boosting the Transferability of Adversarial Attacks with Reverse Adversarial Perturbation (22 citations)
- Natural Color Fool: Towards Boosting Black-box Unrestricted Attacks (15 citations)
- Adversarial Attack on Attackers: Post-Process to Mitigate Black-Box Score-Based Query Attacks (10 citations)
- Blackbox Attacks via Surrogate Ensemble Search (10 citations)
- MORA: Improving Ensemble Robustness Evaluation with Model Reweighing Attack (5 citations)
- Decision-based Black-box Attack Against Vision Transformers via Patch-wise Adversarial Removal (2 citations)
- Towards Lightweight Black-Box Attack Against Deep Neural Networks (1 citation)
