ICLR 2019 accepted papers on Adversarial robustness and security
29 of the 502 papers accepted at ICLR 2019 (5.8%) are on Adversarial robustness and security, compared with 6.2% at ICLR 2018. They fall into 3 topics.
See ICLR 2019 in the interactive map
Submitting to ICLR? 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 · 22 papers
- Robustness May Be at Odds with Accuracy (356 citations)
- Evaluating Robustness of Neural Networks with Mixed Integer Programming (328 citations)
- Benchmarking Neural Network Robustness to Common Corruptions and Perturbations (299 citations)
- Towards the first adversarially robust neural network model on MNIST (237 citations)
- Structured Adversarial Attack: Towards General Implementation and Better Interpretability (114 citations)
- Training for Faster Adversarial Robustness Verification via Inducing ReLU Stability (107 citations)
- Adv-BNN: Improved Adversarial Defense through Robust Bayesian Neural Network (99 citations)
- Hierarchical interpretations for neural network predictions (97 citations)
- Boosting Robustness Certification of Neural Networks (91 citations)
- The Limitations of Adversarial Training and the Blind-Spot Attack (90 citations)
- Improving the Generalization of Adversarial Training with Domain Adaptation (84 citations)
- Are adversarial examples inevitable? (80 citations)
and 10 more in the interactive map.
Attacks on text and graphs · 4 papers
- Adversarial Attacks on Graph Neural Networks via Meta Learning (373 citations)
- Characterizing Audio Adversarial Examples Using Temporal Dependency (98 citations)
- Beyond Pixel Norm-Balls: Parametric Adversaries using an Analytically Differentiable Renderer (28 citations)
- Don't let your Discriminator be fooled (21 citations)
