Learning Conformal Abstention Policies for Adaptive Risk Management in Large Language and Vision-Language Models
Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel pith:IHLKM47Zrecord.jsonopen to challenge →
read the original abstract
Large Language and Vision-Language Models (LLMs/VLMs) are increasingly used in safety-critical applications, yet their opaque decision-making complicates risk assessment and reliability. Uncertainty quantification (UQ) helps assess prediction confidence and enables abstention when uncertainty is high. Conformal prediction (CP), a leading UQ method, provides statistical guarantees but relies on static thresholds, which fail to adapt to task complexity and evolving data distributions, leading to suboptimal trade-offs in accuracy, coverage, and informativeness. To address this, we propose learnable conformal abstention, integrating reinforcement learning (RL) with CP to optimize abstention thresholds dynamically. By treating CP thresholds as adaptive actions, our approach balances multiple objectives, minimizing prediction set size while maintaining reliable coverage. Extensive evaluations across diverse LLM/VLM benchmarks show our method outperforms Least Ambiguous Classifiers (LAC) and Adaptive Prediction Sets (APS), improving accuracy by up to 3.2%, boosting AUROC for hallucination detection by 22.19%, enhancing uncertainty-guided selective generation (AUARC) by 21.17%, and reducing calibration error by 70%-85%. These improvements hold across multiple models and datasets while consistently meeting the 90% coverage target, establishing our approach as a more effective and flexible solution for reliable decision-making in safety-critical applications. The code is available at: {https://github.com/sinatayebati/vlm-uncertainty}.
This paper has not been read by Pith yet.
Forward citations
Cited by 8 Pith papers
-
VLM Judges Can Rank but Cannot Score: Task-Dependent Uncertainty in Multimodal Evaluation
VLM judges exhibit task-dependent uncertainty in their scores, with conformal prediction revealing wide intervals for complex tasks and a decoupling between good ranking performance and poor absolute scoring reliability.
-
From Scalars to Tensors: Declared Losses Recover Epistemic Distinctions That Neutrosophic Scalars Cannot Express
Declared losses recover epistemic distinctions collapsed by scalar neutrosophic T/I/F values in LLM evaluations.
-
Uncertainty Quantification for Computer-Use Agents: A Benchmark across Vision-Language Models and GUI Grounding Datasets
Argus benchmark shows UQ method rankings for GUI grounding agents are stable within models across datasets but degrade across model classes and to closed-source vendors.
-
GroundControl: Anticipating Navigation Failures in Vision-Language Agents via Trajectory-Consistent Uncertainty Estimates
GroundControl estimates navigation uncertainty via statistical deviation from nominal goal-directed distance dynamics, achieving low E-AURC (0.0024 weighted for GPT-4o) on EB-Navigation splits and outperforming entrop...
-
Learning When to Remember: Risk-Sensitive Contextual Bandits for Abstention-Aware Memory Retrieval in LLM-Based Coding Agents
RSCB-MC is a risk-sensitive contextual bandit memory controller for LLM coding agents that chooses safe actions including abstention, achieving 60.5% proxy success with 0% false positives and low latency in 200-case v...
-
Strategic Decision Support for AI Agents
The paper introduces an optimization framework for AI agents to strategically seek support, proving a threshold policy on support value and providing an online algorithm to control missed-support error without distrib...
-
LLMs Uncertainty Quantification via Adaptive Conformal Semantic Entropy
ACSE estimates LLM prompt uncertainty via adaptive clustering of semantic entropy across multiple responses and uses conformal prediction to bound error rates on accepted answers with distribution-free guarantees.
-
LLMs Uncertainty Quantification via Adaptive Conformal Semantic Entropy
ACSE estimates LLM uncertainty via adaptive semantic entropy clustering with conformal prediction guarantees, reporting higher AUROC than token entropy baselines on datasets like TriviaQA.
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.