# Refine and Align: Confidence Calibration through Multi-Agent Interaction in VQA

arXiv: https://arxiv.org/abs/2511.11169 · HTML: https://mlsanity.com/paper/2511.11169/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Placed in Multimodal and embodied agents, one of 16 directions in LLM agents and tool use.
- Builds on 28 of its 55 references that are in the map.
- First posted on arXiv 14 Nov 2025.

## Placed in

- LLM agents and tool use / [Multimodal and embodied agents](https://mlsanity.com/agents/multimodal-and-embodied-agents/)

## Builds on (most cited first, 28 of 28)

- [Qwen2.5-VL Technical Report](https://mlsanity.com/paper/2502.13923/) (2025)
- [LLaVA-OneVision: Easy Visual Task Transfer](https://mlsanity.com/paper/2408.03326/) (2024)
- [VQA: Visual Question Answering](https://mlsanity.com/paper/1505.00468/) (2015)
- [Gemma 3 Technical Report](https://mlsanity.com/paper/2503.19786/) (2025)
- [Measuring and Narrowing the Compositionality Gap in Language Models](https://mlsanity.com/paper/2210.03350/) (2022)
- [VizWiz Grand Challenge: Answering Visual Questions from Blind People](https://mlsanity.com/paper/1802.08218/) (2018)
- [Just Ask for Calibration: Strategies for Eliciting Calibrated Confidence Scores from Language Models Fine-Tuned with Human Feedback](https://mlsanity.com/paper/2305.14975/) (2023)
- [DriveLM: Driving with Graph Visual Question Answering](https://mlsanity.com/paper/2312.14150/) (2023)
- [Phi-4-Mini Technical Report: Compact yet Powerful Multimodal Language Models via Mixture-of-LoRAs](https://mlsanity.com/paper/2503.01743/) (2025)
- [Generate rather than Retrieve: Large Language Models are Strong Context Generators](https://mlsanity.com/paper/2209.10063/) (2022)
- [NuScenes-QA: A Multi-modal Visual Question Answering Benchmark for Autonomous Driving Scenario](https://mlsanity.com/paper/2305.14836/) (2023)
- [LingoQA: Visual Question Answering for Autonomous Driving](https://mlsanity.com/paper/2312.14115/) (2023)
- [Overconfidence is Key: Verbalized Uncertainty Evaluation in Large Language and Vision-Language Models](https://mlsanity.com/paper/2405.02917/) (2024)
- [Does Thinking More always Help? Mirage of Test-Time Scaling in Reasoning Models](https://mlsanity.com/paper/2506.04210/) (2025)
- [Meta-Calibration: Learning of Model Calibration Using Differentiable Expected Calibration Error](https://mlsanity.com/paper/2106.09613/) (2021)
- [Confidence Calibration and Rationalization for LLMs via Multi-Agent Deliberation](https://mlsanity.com/paper/2404.09127/) (2024)
- [Class-Distribution-Aware Calibration for Long-Tailed Visual Recognition](https://mlsanity.com/paper/2109.05263/) (2021)
- [MoCaE: Mixture of Calibrated Experts Significantly Improves Object Detection](https://mlsanity.com/paper/2309.14976/) (2023)
- [Explaining Autonomous Driving Actions with Visual Question Answering](https://mlsanity.com/paper/2307.10408/) (2023)
- [Multi-Agent VQA: Exploring Multi-Agent Foundation Models in Zero-Shot Visual Question Answering](https://mlsanity.com/paper/2403.14783/) (2024)
- [Towards Top-Down Reasoning: An Explainable Multi-Agent Approach for Visual Question Answering](https://mlsanity.com/paper/2311.17331/) (2023)
- [Visual Question Answering in the Medical Domain](https://mlsanity.com/paper/2309.11080/) (2023)
- [Multi-Agent Debate Strategies to Enhance Requirements Engineering with Large Language Models](https://mlsanity.com/paper/2507.05981/) (2025)
- [Selectively Answering Visual Questions](https://mlsanity.com/paper/2406.00980/) (2024)
- [Variational Visual Question Answering for Uncertainty-Aware Selective Prediction](https://mlsanity.com/paper/2505.09591/) (2025)
- [Multi-Agents Based on Large Language Models for Knowledge-based Visual Question Answering](https://mlsanity.com/paper/2412.18351/) (2024)
- [Medical visual question answering using joint self-supervised learning](https://mlsanity.com/paper/2302.13069/) (2023)
- [Classifier Ensemble for Efficient Uncertainty Calibration of Deep Neural Networks for Image Classification](https://mlsanity.com/paper/2501.10089/) (2025)

## Built on by (most cited first, 4 of 4)

- [Overconfidence and Calibration in Medical VQA: Empirical Findings and Hallucination-Aware Mitigation](https://mlsanity.com/paper/2604.02543/) (2026)
- [Seeing Before Agreeing: Aligning Multi-Agent Consensus with Visual Evidence](https://mlsanity.com/paper/2605.30698/) (2026)
- [Confidence Composition for Multiagent Language Model Systems](https://mlsanity.com/paper/2606.13591/) (2026)
- [Uncertainty-Aware Decision Making in Multimodal Large Language Models](https://mlsanity.com/paper/2608.17084/) (2026)

## Abstract (arXiv)

In the context of Visual Question Answering (VQA) and Agentic AI, calibration refers to how closely an AI system's confidence in its answers reflects their actual correctness. This aspect becomes especially important when such systems operate autonomously and must make decisions under visual uncertainty. While modern VQA systems, powered by advanced vision-language models (VLMs), are increasingly used in high-stakes domains like medical diagnostics and autonomous navigation due to their improved accuracy, the reliability of their confidence estimates remains under-examined. Particularly, these systems often produce overconfident responses. To address this, we introduce AlignVQA, a debate-based multi-agent framework, in which diverse specialized VLM -- each following distinct prompting strategies -- generate candidate answers and then engage in two-stage interaction: generalist agents critique, refine and aggregate these proposals. This debate process yields confidence estimates that more accurately reflect the model's true predictive performance. We find that more calibrated specialized agents produce better aligned confidences. Furthermore, we introduce a novel differentiable calibration-aware loss function called aligncal designed to fine-tune the specialized agents by minimizing an upper bound on the calibration error. This objective explicitly improves the fidelity of each agent's confidence estimates. Empirical results across multiple benchmark VQA datasets substantiate the efficacy of our approach, demonstrating substantial reductions in calibration discrepancies. Furthermore, we propose a novel differentiable calibration-aware loss to fine-tune the specialized agents and improve the quality of their individual confidence estimates based on minimising upper bound calibration error.