AI Safety

Robust Safety under Stochastic Uncertainty with Discrete-Time Control Barrier Functions RSS 2023 ROSCOE: A Suite of Metrics for Scoring Step-by-Step Reasoning ICLR 2023 RS-Del: Edit Distance Robustness Certificates for Sequence Classifiers via Randomized Deletion NeurIPS 2023 Run-off Election: Improved Provable Defense against Data Poisoning Attacks ICML 2023 SafeConv: Explaining and Correcting Conversational Unsafe Behavior ACL 2023 Safe Exploration in Reinforcement Learning: A Generalized Formulation and Algorithms NeurIPS 2023 Safe Interactive Autonomy for Multi-Agent Systems AAAI 2023 Safe Reinforcement Learning via Probabilistic Logic Shields IJCAI 2023 Safe Reinforcement Learning via Shielding under Partial Observability AAAI 2023 Safety Gymnasium: A Unified Safe Reinforcement Learning Benchmark NeurIPS 2023 Safety Validation of Learning-Based Autonomous Systems: A Multi-Fidelity Approach AAAI 2023 Safety Verification and Universal Invariants for Relational Action Bases IJCAI 2023 Safety Verification of Decision-Tree Policies in Continuous Time NeurIPS 2023 Sample-Efficient and Safe Deep Reinforcement Learning via Reset Deep Ensemble Agents NeurIPS 2023 Sample Efficient Paradigms for Personalized Assessment of Taskable AI Systems IJCAI 2023 Scalable Edge Blocking Algorithms for Defending Active Directory Style Attack Graphs AAAI 2023 SCALE-UP: An Efficient Black-box Input-level Backdoor Detection via Analyzing Scaled Prediction Consistency ICLR 2023 Scaling Laws for Reward Model Overoptimization ICML 2023 Security Challenges in Natural Language Processing Models EMNLP 2023 Seeing is not Believing: Robust Reinforcement Learning against Spurious Correlation NeurIPS 2023 SelfCheckGPT: Zero-Resource Black-Box Hallucination Detection for Generative Large Language Models EMNLP 2023 Self-Criticism: Aligning Large Language Models with their Understanding of Helpfulness, Honesty, and Harmlessness EMNLP 2023 Self-Detoxifying Language Models via Toxification Reversal EMNLP 2023 Sequential Strategic Screening ICML 2023 Setting the Trap: Capturing and Defeating Backdoors in Pretrained Language Models through Honeypots NeurIPS 2023