ICCV 2019 accepted papers on Adversarial robustness and security
30 of the 1,075 papers accepted at ICCV 2019 (2.8%) are on Adversarial robustness and security, compared with 0.8% at ICCV 2017. They fall into 3 topics.
See ICCV 2019 in the interactive map
Submitting to ICCV? 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 · 12 papers
- Bit-Flip Attack: Crushing Neural Network With Progressive Bit Search (298 citations)
- Towards Adversarially Robust Object Detection (163 citations)
- Universal Perturbation Attack Against Image Retrieval (120 citations)
- Semantic Adversarial Attacks: Parametric Transformations That Fool Deep Classifiers (96 citations)
- Attacking Optical Flow (91 citations)
- Physical Adversarial Textures That Fool Visual Object Tracking (84 citations)
- Targeted Mismatch Adversarial Attack: Query With a Flower to Retrieve the Tower (72 citations)
- Fooling Network Interpretation in Image Classification (68 citations)
- advPattern: Physical-World Attacks on Deep Person Re-Identification via Adversarially Transformable Patterns (59 citations)
- CIIDefence: Defeating Adversarial Attacks by Fusing Class-Specific Image Inpainting and Image Denoising (58 citations)
- Hilbert-Based Generative Defense for Adversarial Examples (51 citations)
- AdvIT: Adversarial Frames Identifier Based on Temporal Consistency in Videos (45 citations)
Certified and empirical robustness · 10 papers
- Human Uncertainty Makes Classification More Robust (233 citations)
- Sparse and Imperceivable Adversarial Attacks (206 citations)
- Scalable Verified Training for Provably Robust Image Classification (161 citations)
- Adversarial Robustness vs. Model Compression, or Both? (149 citations)
- Bilateral Adversarial Training: Towards Fast Training of More Robust Models Against Adversarial Attacks (118 citations)
- Universal Adversarial Perturbation via Prior Driven Uncertainty Approximation (98 citations)
- Improving Adversarial Robustness via Guided Complement Entropy (57 citations)
- Defending Against Universal Perturbations With Shared Adversarial Training (56 citations)
- Adversarial Learning With Margin-Based Triplet Embedding Regularization (48 citations)
- On the Global Optima of Kernelized Adversarial Representation Learning (22 citations)
Black-box and transfer attacks · 8 papers
- Enhancing Adversarial Example Transferability With an Intermediate Level Attack (276 citations)
- Adversarial Defense by Restricting the Hidden Space of Deep Neural Networks (158 citations)
- Guessing Smart: Biased Sampling for Efficient Black-Box Adversarial Attacks (124 citations)
- Adversarial Defense via Learning to Generate Diverse Attacks (114 citations)
- On the Design of Black-Box Adversarial Examples by Leveraging Gradient-Free Optimization and Operator Splitting Method (63 citations)
- A Geometry-Inspired Decision-Based Attack (46 citations)
- Once a MAN: Towards Multi-Target Attack via Learning Multi-Target Adversarial Network Once (33 citations)
- The LogBarrier Adversarial Attack: Making Effective Use of Decision Boundary Information (29 citations)
