AI Safety

Lexicographic Multi-Objective Reinforcement Learning IJCAI 2022 Linearity Grafting: Relaxed Neuron Pruning Helps Certifiable Robustness ICML 2022 LOT: Layer-wise Orthogonal Training on Improving l2 Certified Robustness NeurIPS 2022 Measuring Data Reconstruction Defenses in Collaborative Inference Systems NeurIPS 2022 Mitigating Covertly Unsafe Text within Natural Language Systems EMNLP 2022 Mitigating Toxic Degeneration with Empathetic Data: Exploring the Relationship Between Toxicity and Empathy NAACL 2022 Model AI Assignments 2022 AAAI 2022 Modeling Adversarial Noise for Adversarial Training ICML 2022 Moderate-fitting as a Natural Backdoor Defender for Pre-trained Language Models NeurIPS 2022 MORA: Improving Ensemble Robustness Evaluation with Model Reweighing Attack NeurIPS 2022 NASPY: Automated Extraction of Automated Machine Learning Models ICLR 2022 NaturalAdversaries: Can Naturalistic Adversaries Be as Effective as Artificial Adversaries? EMNLP 2022 Neural Abstractions NeurIPS 2022 Neural Theory-of-Mind? On the Limits of Social Intelligence in Large LMs EMNLP 2022 Neuro-Symbolic Verification of Deep Neural Networks IJCAI 2022 NICGSlowDown: Evaluating the Efficiency Robustness of Neural Image Caption Generation Models CVPR 2022 Non-Transferable Learning: A New Approach for Model Ownership Verification and Applicability Authorization ICLR 2022 Not All Poisons are Created Equal: Robust Training against Data Poisoning ICML 2022 On Collective Robustness of Bagging Against Data Poisoning ICML 2022 One-shot Neural Backdoor Erasing via Adversarial Weight Masking NeurIPS 2022 On Global-view Based Defense via Adversarial Attack and Defense Risk Guaranteed Bounds AISTATS 2022 Online Adversarial Attacks ICLR 2022 On the Adversarial Robustness of Causal Algorithmic Recourse ICML 2022 On the Effectiveness of Small Input Noise for Defending Against Query-Based Black-Box Attacks WACV 2022 On the Impact of Spurious Correlation for Out-of-Distribution Detection AAAI 2022