Exploring Response Uncertainty in MLLMs: An Empirical Evaluation under Misleading Scenarios

• URL: http://arxiv.org/abs/2411.02708v3
• Date: Wed, 03 Sep 2025 02:51:26 GMT
• Title: Exploring Response Uncertainty in MLLMs: An Empirical Evaluation under Misleading Scenarios
• Authors: Yunkai Dang, Mengxi Gao, Yibo Yan, Xin Zou, Yanggan Gu, Jungang Li, Jingyu Wang, Peijie Jiang, Aiwei Liu, Jia Liu, Xuming Hu,
• Abstract summary: Multimodal large language models (MLLMs) have recently achieved state-of-the-art performance on tasks ranging from visual question answering to video understanding.<n>We reveal a response uncertainty phenomenon: twelve state-of-the-art open-source MLLMs overturn a previously correct answer in 65% of cases after receiving a single deceptive cue.
• Score: 49.53589774730807
• License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/
• Abstract: Multimodal large language models (MLLMs) have recently achieved state-of-the-art performance on tasks ranging from visual question answering to video understanding. However, existing studies have concentrated mainly on visual-textual misalignment, leaving largely unexplored the MLLMs' ability to preserve an originally correct answer when confronted with misleading information. We reveal a response uncertainty phenomenon: across nine standard datasets, twelve state-of-the-art open-source MLLMs overturn a previously correct answer in 65% of cases after receiving a single deceptive cue. To systematically quantify this vulnerability, we propose a two-stage evaluation pipeline: (1) elicit each model's original response on unperturbed inputs; (2) inject explicit (false-answer hints) and implicit (contextual contradictions) misleading instructions, and compute the misleading rate - the fraction of correct-to-incorrect flips. Leveraging the most susceptible examples, we curate the Multimodal Uncertainty Benchmark (MUB), a collection of image-question pairs stratified into low, medium, and high difficulty based on how many of twelve state-of-the-art MLLMs they mislead. Extensive evaluation on twelve open-source and five closed-source models reveals a high uncertainty: average misleading rates exceed 86%, with explicit cues over 67.19% and implicit cues over 80.67%. To reduce the misleading rate, we then fine-tune all open-source MLLMs on a compact 2000-sample mixed-instruction dataset, reducing misleading rates to 6.97% (explicit) and 32.77% (implicit), boosting consistency by nearly 29.37% on highly deceptive inputs, and slightly improving accuracy on standard benchmarks. Our code is available at https://github.com/Yunkaidang/uncertainty

Related papers

• Bias Testing and Mitigation in Black Box LLMs using Metamorphic Relations [12.095552079739933]
  We propose a unified framework for both systematic bias evaluation and targeted mitigation.<n>Our approach introduces six novel Metamorphic Relations (MRs)<n>MRs transform direct bias-inducing inputs into semantically equivalent yet adversarially challenging variants.
  arXiv  Detail & Related papers  (2025-11-29T16:56:38Z)
• CLUE: Non-parametric Verification from Experience via Hidden-State Clustering [64.50919789875233]
  We show that correctness of a solution is encoded as a geometrically separable signature within the trajectory of hidden activations.<n>ClUE consistently outperforms LLM-as-a-judge baselines and matches or exceeds modern confidence-based methods in reranking candidates.
  arXiv  Detail & Related papers  (2025-10-02T02:14:33Z)
• Clotho: Measuring Task-Specific Pre-Generation Test Adequacy for LLM Inputs [6.862079218077768]
  Testing Large Language Models on specific tasks is difficult and costly.<n>A key challenge is to assess input adequacy in a way that reflects the demands of the task.<n>We introduce CLOTHO, a task-specific, pre-generation adequacy measure.
  arXiv  Detail & Related papers  (2025-09-22T02:34:09Z)
• Protecting multimodal large language models against misleading visualizations [94.71976205962527]
  We introduce the first inference-time methods to improve performance on misleading visualizations. We find that MLLM question-answering accuracy drops on average to the level of a random baseline.
  arXiv  Detail & Related papers  (2025-02-27T20:22:34Z)
• None of the Others: a General Technique to Distinguish Reasoning from Memorization in Multiple-Choice LLM Evaluation Benchmarks [0.9831489366502301]
  We introduce a general variation method for multiple-choice questions that completely dissociates the correct answer from previously seen tokens or concepts.<n>Using this method, we evaluate state-of-the-art proprietary and open-source LLMs on two datasets available in English and Spanish.<n>Results show that all models experience remarkable accuracy drops, with an average loss of 57% on MMLU and 50% on UNED-Access 2024.
  arXiv  Detail & Related papers  (2025-02-18T14:32:44Z)
• Aligning Large Language Models for Faithful Integrity Against Opposing Argument [71.33552795870544]
  Large Language Models (LLMs) have demonstrated impressive capabilities in complex reasoning tasks. They can be easily misled by unfaithful arguments during conversations, even when their original statements are correct. We propose a novel framework, named Alignment for Faithful Integrity with Confidence Estimation.
  arXiv  Detail & Related papers  (2025-01-02T16:38:21Z)
• Drawing the Line: Enhancing Trustworthiness of MLLMs Through the Power of Refusal [21.342265570934995]
  Existing methods have largely overlooked the importance of refusal responses as a means of enhancing MLLMs reliability. We present the Information Boundary-aware Learning Framework (InBoL), a novel approach that empowers MLLMs to refuse to answer user queries when encountering insufficient information. This framework introduces a comprehensive data generation pipeline and tailored training strategies to improve the model's ability to deliver appropriate refusal responses.
  arXiv  Detail & Related papers  (2024-12-15T14:17:14Z)
• mR$^2$AG: Multimodal Retrieval-Reflection-Augmented Generation for Knowledge-Based VQA [78.45521005703958]
  multimodal Retrieval-Augmented Generation (mRAG) is naturally introduced to provide MLLMs with comprehensive and up-to-date knowledge. We propose a novel framework called textbfRetrieval-textbfReftextbfAugmented textbfGeneration (mR$2$AG) which achieves adaptive retrieval and useful information localization. mR$2$AG significantly outperforms state-of-the-art MLLMs on INFOSEEK and Encyclopedic-VQA
  arXiv  Detail & Related papers  (2024-11-22T16:15:50Z)
• MQM-APE: Toward High-Quality Error Annotation Predictors with Automatic Post-Editing in LLM Translation Evaluators [53.91199933655421]
  Large Language Models (LLMs) have shown significant potential as judges for Machine Translation (MT) quality assessment. We introduce a universal and training-free framework, $textbfMQM-APE, to enhance the quality of error annotations predicted by LLM evaluators.
  arXiv  Detail & Related papers  (2024-09-22T06:43:40Z)
• DefAn: Definitive Answer Dataset for LLMs Hallucination Evaluation [39.857198257988685]
  Large Language Models (LLMs) have demonstrated remarkable capabilities, revolutionizing the integration of AI in daily life applications. They are prone to hallucinations, generating claims that contradict established facts, and producing inconsistent responses when the same prompt is presented multiple times. This paper introduces a comprehensive benchmark dataset comprising over 75,000 prompts across eight domains.
  arXiv  Detail & Related papers  (2024-06-13T14:18:13Z)
• CaLM: Contrasting Large and Small Language Models to Verify Grounded Generation [76.31621715032558]
  Grounded generation aims to equip language models (LMs) with the ability to produce more credible and accountable responses. We introduce CaLM, a novel verification framework. Our framework empowers smaller LMs, which rely less on parametric memory, to validate the output of larger LMs.
  arXiv  Detail & Related papers  (2024-06-08T06:04:55Z)
• Cycles of Thought: Measuring LLM Confidence through Stable Explanations [53.15438489398938]
  Large language models (LLMs) can reach and even surpass human-level accuracy on a variety of benchmarks, but their overconfidence in incorrect responses is still a well-documented failure mode. We propose a framework for measuring an LLM's uncertainty with respect to the distribution of generated explanations for an answer.
  arXiv  Detail & Related papers  (2024-06-05T16:35:30Z)
• Semantic Density: Uncertainty Quantification for Large Language Models through Confidence Measurement in Semantic Space [14.715989394285238]
  Existing Large Language Models (LLMs) do not have an inherent functionality to provide the users with an uncertainty/confidence metric for each response it generates. A new framework is proposed in this paper to address these issues. Semantic density extracts uncertainty/confidence information for each response from a probability distribution perspective in semantic space.
  arXiv  Detail & Related papers  (2024-05-22T17:13:49Z)
• CyberSecEval 2: A Wide-Ranging Cybersecurity Evaluation Suite for Large Language Models [6.931433424951554]
  Large language models (LLMs) introduce new security risks, but there are few comprehensive evaluation suites to measure and reduce these risks. We present BenchmarkName, a novel benchmark to quantify LLM security risks and capabilities. We evaluate multiple state-of-the-art (SOTA) LLMs, including GPT-4, Mistral, Meta Llama 3 70B-Instruct, and Code Llama.
  arXiv  Detail & Related papers  (2024-04-19T20:11:12Z)
• Multicalibration for Confidence Scoring in LLMs [6.948522445499497]
  This paper proposes the use of "multicalibration" to yield interpretable and reliable confidence scores for outputs generated by large language models (LLMs) We show how to form groupings for prompt/completion pairs that are correlated with the probability of correctness via two techniques: clustering within an embedding space, and "self-annotation" We show how our techniques can yield confidence scores that provide substantial improvements in fine-grained measures of both calibration and accuracy compared to existing methods.
  arXiv  Detail & Related papers  (2024-04-06T17:33:37Z)
• Are We on the Right Way for Evaluating Large Vision-Language Models? [92.5761176224556]
  Large vision-language models (LVLMs) have recently achieved rapid progress, sparking numerous studies to evaluate their multi-modal capabilities. We identify two primary issues: Visual content is unnecessary for many samples and intentional data leakage exists. We present MMStar, an elite vision-indispensable multi-modal benchmark comprising 1,500 samples meticulously selected by humans.
  arXiv  Detail & Related papers  (2024-03-29T17:59:34Z)
• How Easy is It to Fool Your Multimodal LLMs? An Empirical Analysis on Deceptive Prompts [54.07541591018305]
  We present MAD-Bench, a benchmark that contains 1000 test samples divided into 5 categories, such as non-existent objects, count of objects, and spatial relationship. We provide a comprehensive analysis of popular MLLMs, ranging from GPT-4v, Reka, Gemini-Pro, to open-sourced models, such as LLaVA-NeXT and MiniCPM-Llama3. While GPT-4o achieves 82.82% accuracy on MAD-Bench, the accuracy of any other model in our experiments ranges from 9% to 50%.
  arXiv  Detail & Related papers  (2024-02-20T18:31:27Z)
• Fake Alignment: Are LLMs Really Aligned Well? [91.26543768665778]
  This study investigates the substantial discrepancy in performance between multiple-choice questions and open-ended questions. Inspired by research on jailbreak attack patterns, we argue this is caused by mismatched generalization.
  arXiv  Detail & Related papers  (2023-11-10T08:01:23Z)
• Assessing the Reliability of Large Language Model Knowledge [78.38870272050106]
  Large language models (LLMs) have been treated as knowledge bases due to their strong performance in knowledge probing tasks. How do we evaluate the capabilities of LLMs to consistently produce factually correct answers? We propose MOdel kNowledge relIabiliTy scORe (MONITOR), a novel metric designed to directly measure LLMs' factual reliability.
  arXiv  Detail & Related papers  (2023-10-15T12:40:30Z)
• Quantifying Uncertainty in Answers from any Language Model and Enhancing their Trustworthiness [16.35655151252159]
  We introduce BSDetector, a method for detecting bad and speculative answers from a pretrained Large Language Model. Our uncertainty quantification technique works for any LLM accessible only via a black-box API.
  arXiv  Detail & Related papers  (2023-08-30T17:53:25Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.