Online Conformal Abstention Under Adversarial Bandit Feedback

arXiv:2506.14067 · cs.LG · Submitted 2025-06-16 · Read on arXiv

cs.LG

Submitted: 2025-06-16

Updated: 2026-10-03

Code: https://github.com/langchain-ai/langchain

Terminology

• Abstention — 4× in this paper · explained in 1 episode
• Regret — 2× in this paper · explained in 3 episodes
• Large Language Models (LLMs) — 1× in this paper · explained in 6 episodes

Sources

• Learn then Test: Calibrating Predictive Algorithms to Achieve Risk Control
• Foundations of Reinforcement Learning and Interactive Decision Making
• Stochastic Online Conformal Prediction with Semi-Bandit Feedback
• Hierarchical Selective Classification
• Language Models (Mostly) Know What They Know
• Selective Question Answering under Domain Shift
• Learning to Reject with a Fixed Predictor: Application to Decontextualization
• SQuAD: 100,000+ Questions for Machine Comprehension of Text
• Exponentially Weighted Algorithm for Online Network Resource Allocation with Long-Term Constraints
• Self-Consistency Improves Chain of Thought Reasoning in Language Models
• Mitigating LLM Hallucinations via Conformal Abstention
• Online Conformal Prediction with Adversarial Semi-bandit Feedback via Regret Minimization

Related papers

• Polynomial-Augmented Neural Networks (PANNs) with Weak Orthogonality Constraints for Enhanced Function and PDE Approximation
• AIRL-S: Unifying Reinforcement Learning and Search-Based Test-Time Scaling via Adversarial Inverse Reinforcement Learning
• Transformers as Bayesian In-Context Experimenters: Smoothness-Adaptive Efficient ATE Estimation
• Convergence issues in Relational Concept Analysis based on AOC-posets
• Beliefs Beyond Posteriors: The Predictive Cost of ELBO Binding
• Understanding Diffusion Models via Ratio-Based Function Approximation with SignReLU Networks