ICML 2020 accepted papers on Adversarial robustness and security
49 of the 1,087 papers accepted at ICML 2020 (4.5%) are on Adversarial robustness and security, compared with 5.2% at ICML 2019. They fall into 3 topics.
See ICML 2020 in the interactive map
Submitting to ICML? 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 · 40 papers
- Reliable evaluation of adversarial robustness with an ensemble of diverse parameter-free attacks (582 citations)
- Adversarial Filters of Dataset Biases (159 citations)
- Minimally distorted Adversarial Examples with a Fast Adaptive Boundary Attack (118 citations)
- Attacks Which Do Not Kill Training Make Adversarial Learning Stronger (83 citations)
- Puzzle Mix: Exploiting Saliency and Local Statistics for Optimal Mixup (71 citations)
- Adversarial Robustness Against the Union of Multiple Perturbation Models (69 citations)
- Randomized Smoothing of All Shapes and Sizes (66 citations)
- Confidence-Calibrated Adversarial Training: Generalizing to Unseen Attacks (61 citations)
- Fundamental Tradeoffs between Invariance and Sensitivity to Adversarial Perturbations (52 citations)
- Certified Robustness to Label-Flipping Attacks via Randomized Smoothing (48 citations)
- Understanding and Mitigating the Tradeoff between Robustness and Accuracy (45 citations)
- Overfitting in adversarially robust deep learning (43 citations)
and 28 more in the interactive map.
Black-box and transfer attacks · 5 papers
- Transfer Learning without Knowing: Reprogramming Black-box Machine Learning Models with Scarce Data and Limited Resources (27 citations)
- Adversarial Attacks on Copyright Detection Systems (25 citations)
- Robust and Stable Black Box Explanations (5 citations)
- Black-box Certification and Learning under Adversarial Perturbations (4 citations)
- Dual-Path Distillation: A Unified Framework to Improve Black-Box Attacks (4 citations)
Attacks on text and graphs · 4 papers
- Informative Dropout for Robust Representation Learning: A Shape-bias Perspective (42 citations)
- Rank Aggregation from Pairwise Comparisons in the Presence of Adversarial Corruptions (5 citations)
- On Breaking Deep Generative Model-based Defenses and Beyond (2 citations)
- Robustness to Programmable String Transformations via Augmented Abstract Training
