AAAI 2020 accepted papers on Adversarial robustness and security
49 of the 1,610 papers accepted at AAAI 2020 (3.0%) are on Adversarial robustness and security, compared with 1.8% at AAAI 2019. They fall into 4 topics.
See AAAI 2020 in the interactive map
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 · 18 papers
- Is BERT Really Robust? A Strong Baseline for Natural Language Attack on Text Classification and Entailment (1,220 citations)
- Stealthy and Efficient Adversarial Attacks against Deep Reinforcement Learning (127 citations)
- ML-LOO: Detecting Adversarial Examples with Feature Attribution (106 citations)
- Adversarial Learning of Privacy-Preserving and Task-Oriented Representations (73 citations)
- Spatiotemporally Constrained Action Space Attacks on Deep Reinforcement Learning Agents (62 citations)
- CD-UAP: Class Discriminative Universal Adversarial Perturbation (59 citations)
- Weighted-Sampling Audio Adversarial Example Attack (47 citations)
- Revisiting the Foundations of Abstract Argumentation – Semantics Based on Weak Admissibility and Weak Defense (46 citations)
- Learning Efficient Representations for Fake Speech Detection (41 citations)
- Joint Character-Level Word Embedding and Adversarial Stability Training to Defend Adversarial Text (36 citations)
- Beyond Digital Domain: Fooling Deep Learning Based Recognition System in Physical World (32 citations)
- ECGadv: Generating Adversarial Electrocardiogram to Misguide Arrhythmia Classification System (24 citations)
and 6 more in the interactive map.
Certified and empirical robustness · 16 papers
- Seq2Sick: Evaluating the Robustness of Sequence-to-Sequence Models with Adversarial Examples (233 citations)
- Adversarially Robust Distillation (227 citations)
- Universal Adversarial Training (168 citations)
- Generating Adversarial Examples for Holding Robustness of Source Code Processing Models (128 citations)
- Robustness Certificates for Sparse Adversarial Attacks by Randomized Ablation (69 citations)
- Abstract Interpretation of Decision Tree Ensemble Classifiers (47 citations)
- Fastened CROWN: Tightened Neural Network Robustness Certificates (40 citations)
- A Robust Adversarial Training Approach to Machine Reading Comprehension (38 citations)
- Towards Certificated Model Robustness Against Weight Perturbations (38 citations)
- Invariant Representations through Adversarial Forgetting (35 citations)
- Finding Needles in a Moving Haystack: Prioritizing Alerts with Adversarial Reinforcement Learning (24 citations)
- Identifying Model Weakness with Adversarial Examiner (22 citations)
and 4 more in the interactive map.
Black-box and transfer attacks · 14 papers
- A Restricted Black-Box Adversarial Framework Towards Attacking Graph Embedding Models (144 citations)
- ActiveThief: Model Extraction Using Active Learning and Unannotated Public Data (139 citations)
- Learning Transferable Adversarial Examples via Ghost Networks (139 citations)
- Model Watermarking for Image Processing Networks (137 citations)
- Heuristic Black-Box Adversarial Attacks on Video Recognition Models (94 citations)
- A Frank-Wolfe Framework for Efficient and Effective Adversarial Attacks (62 citations)
- Towards Query-Efficient Black-Box Adversary with Zeroth-Order Natural Gradient Descent (45 citations)
- Adversarial Attack on Deep Product Quantization Network for Image Retrieval (39 citations)
- A New Ensemble Adversarial Attack Powered by Long-Term Gradient Memories (30 citations)
- Suspicion-Free Adversarial Attacks on Clustering Algorithms (23 citations)
- SADA: Semantic Adversarial Diagnostic Attacks for Autonomous Applications (21 citations)
- CAG: A Real-Time Low-Cost Enhanced-Robustness High-Transferability Content-Aware Adversarial Attack Generator (18 citations)
and 2 more in the interactive map.
Backdoor and poisoning · 1 paper
- Hidden Trigger Backdoor Attacks (576 citations)
