CVPR 2019 accepted papers on Adversarial robustness and security
35 of the 1,294 papers accepted at CVPR 2019 (2.7%) are on Adversarial robustness and security, compared with 1.8% at CVPR 2018. They fall into 3 topics.
See CVPR 2019 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 · 15 papers
- Multi-Adversarial Discriminative Deep Domain Generalization for Face Presentation Attack Detection (420 citations)
- ComDefend: An Efficient Image Compression Model to Defend Adversarial Examples (326 citations)
- Feature Distillation: DNN-Oriented JPEG Compression Against Adversarial Examples (289 citations)
- Deep Tree Learning for Zero-Shot Face Anti-Spoofing (272 citations)
- SparseFool: A Few Pixels Make a Big Difference (211 citations)
- Adversarial Attacks Beyond the Image Space (141 citations)
- Detection Based Defense Against Adversarial Examples From the Steganalysis Point of View (126 citations)
- Rob-GAN: Generator, Discriminator, and Adversarial Attacker (126 citations)
- Adversarial Defense by Stratified Convolutional Sparse Coding (93 citations)
- Defense Against Adversarial Images Using Web-Scale Nearest-Neighbor Search (57 citations)
- Exact Adversarial Attack to Image Captioning via Structured Output Learning With Latent Variables (54 citations)
- Adversarial Defense Through Network Profiling Based Path Extraction (47 citations)
and 3 more in the interactive map.
Certified and empirical robustness · 11 papers
- Decoupling Direction and Norm for Efficient Gradient-Based L2 Adversarial Attacks and Defenses (255 citations)
- Robustness via Curvature Regularization, and Vice Versa (240 citations)
- Disentangling Adversarial Robustness and Generalization (221 citations)
- Barrage of Random Transforms for Adversarially Robust Defense (158 citations)
- Trust Region Based Adversarial Attack on Neural Networks (63 citations)
- On the Structural Sensitivity of Deep Convolutional Networks to the Directions of Fourier Basis Functions (58 citations)
- A Kernelized Manifold Mapping to Diminish the Effect of Adversarial Perturbations (44 citations)
- ShieldNets: Defending Against Adversarial Attacks Using Probabilistic Adversarial Robustness (36 citations)
- Retrieval-Augmented Convolutional Neural Networks Against Adversarial Examples (8 citations)
- Knowing When to Stop: Evaluation and Verification of Conformity to Output-Size Specifications (6 citations)
- Robustness of 3D Deep Learning in an Adversarial Setting (2 citations)
Black-box and transfer attacks · 9 papers
- Improving Transferability of Adversarial Examples With Input Diversity (1,273 citations)
- Evading Defenses to Transferable Adversarial Examples by Translation-Invariant Attacks (974 citations)
- Feature Denoising for Improving Adversarial Robustness (902 citations)
- Efficient Decision-Based Black-Box Adversarial Attacks on Face Recognition (415 citations)
- Parametric Noise Injection: Trainable Randomness to Improve Deep Neural Network Robustness Against Adversarial Attack (267 citations)
- Feature Space Perturbations Yield More Transferable Adversarial Examples (187 citations)
- MeshAdv: Adversarial Meshes for Visual Recognition (115 citations)
- Curls & Whey: Boosting Black-Box Adversarial Attacks (102 citations)
- Sensitive-Sample Fingerprinting of Deep Neural Networks (86 citations)
