CVPR 2018 accepted papers on Adversarial robustness and security
18 of the 979 papers accepted at CVPR 2018 (1.8%) are on Adversarial robustness and security, compared with 0.1% at CVPR 2017. They fall into 3 topics.
See CVPR 2018 in the interactive map
Submitting to CVPR? 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
Attacks on text and graphs · 9 papers
- Robust Physical-World Attacks on Deep Learning Visual Classification (2,168 citations)
- Crowd Counting via Adversarial Cross-Scale Consistency Pursuit (397 citations)
- Generative Adversarial Perturbations (375 citations)
- On the Robustness of Semantic Segmentation Models to Adversarial Attacks (330 citations)
- Deflecting Adversarial Attacks With Pixel Deflection (327 citations)
- Defense Against Universal Adversarial Perturbations (211 citations)
- NAG: Network for Adversary Generation (144 citations)
- Art of Singular Vectors and Universal Adversarial Perturbations (116 citations)
- Fooling Vision and Language Models Despite Localization and Attention Mechanism (77 citations)
Certified and empirical robustness · 7 papers
- Robust Classification With Convolutional Prototype Learning (473 citations)
- Lightweight Probabilistic Deep Networks (172 citations)
- Empirical Study of the Topology and Geometry of Deep Networks (124 citations)
- Geometric Robustness of Deep Networks: Analysis and Improvement (116 citations)
- Interpret Neural Networks by Identifying Critical Data Routing Paths (100 citations)
- Wasserstein Introspective Neural Networks (63 citations)
- Analytic Expressions for Probabilistic Moments of PL-DNN With Gaussian Input (24 citations)
Black-box and transfer attacks · 2 papers
- Boosting Adversarial Attacks With Momentum (3,047 citations)
- Defense Against Adversarial Attacks Using High-Level Representation Guided Denoiser (945 citations)
