NeurIPS 2020 accepted papers on Adversarial robustness and security
73 of the 1,901 papers accepted at NeurIPS 2020 (3.8%) are on Adversarial robustness and security, compared with 3.6% at NeurIPS 2019. They fall into 4 topics.
See NeurIPS 2020 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 · 49 papers
- Adversarial Weight Perturbation Helps Robust Generalization (204 citations)
- Measuring Robustness to Natural Distribution Shifts in Image Classification (166 citations)
- Automatic Perturbation Analysis for Scalable Certified Robustness and Beyond (112 citations)
- A Closer Look at Accuracy vs. Robustness (99 citations)
- Boosting Adversarial Training with Hypersphere Embedding (89 citations)
- Robust Pre-Training by Adversarial Contrastive Learning (71 citations)
- (De)Randomized Smoothing for Certifiable Defense against Patch Attacks (66 citations)
- Maximum-Entropy Adversarial Data Augmentation for Improved Generalization and Robustness (64 citations)
- Contrastive Learning with Adversarial Examples (61 citations)
- DVERGE: Diversifying Vulnerabilities for Enhanced Robust Generation of Ensembles (60 citations)
- Adversarial Learning for Robust Deep Clustering (46 citations)
- Understanding and Improving Fast Adversarial Training (44 citations)
and 37 more in the interactive map.
Attacks on text and graphs · 12 papers
- On Adaptive Attacks to Adversarial Example Defenses (137 citations)
- Simulating a Primary Visual Cortex at the Front of CNNs Improves Robustness to Image Perturbations (129 citations)
- MetaPoison: Practical General-purpose Clean-label Data Poisoning (92 citations)
- Adversarial Self-Supervised Contrastive Learning (82 citations)
- Watch out! Motion is Blurring the Vision of Your Deep Neural Networks (55 citations)
- Passport-aware Normalization for Deep Model Protection (36 citations)
- Certified Robustness of Graph Convolution Networks for Graph Classification under Topological Attacks (16 citations)
- Adversarial Attacks on Deep Graph Matching (14 citations)
- GreedyFool: Distortion-Aware Sparse Adversarial Attack (12 citations)
- MRI Banding Removal via Adversarial Training (9 citations)
- Targeted Adversarial Perturbations for Monocular Depth Prediction (1 citation)
- Adversarial Bandits with Corruptions
Black-box and transfer attacks · 9 papers
- Diversity can be Transferred: Output Diversification for White- and Black-box Attacks (52 citations)
- Learning Black-Box Attackers with Transferable Priors and Query Feedback (46 citations)
- Denoised Smoothing: A Provable Defense for Pretrained Classifiers (44 citations)
- Perturbing Across the Feature Hierarchy to Improve Standard and Strict Blackbox Attack Transferability (41 citations)
- AdvFlow: Inconspicuous Black-box Adversarial Attacks using Normalizing Flows (19 citations)
- Backpropagating Linearly Improves Transferability of Adversarial Examples (19 citations)
- Towards More Practical Adversarial Attacks on Graph Neural Networks (14 citations)
- An Efficient Adversarial Attack for Tree Ensembles (7 citations)
- Practical No-box Adversarial Attacks against DNNs (6 citations)
Backdoor and poisoning · 3 papers
- Input-Aware Dynamic Backdoor Attack (153 citations)
- Attack of the Tails: Yes, You Really Can Backdoor Federated Learning (109 citations)
- On the Trade-off between Adversarial and Backdoor Robustness (21 citations)
