analiza aiqbench
This paper addresses the critical reliability gap in Multimodal Large Language Models (MLLMs) when operating under adverse conditions, proposing a new benchmark and reasoning framework that explicitly ties cognitive output to visual evidence. By diagnosing failures through a Chain of Evidence approach rather than just final predictions, the work offers a more robust method for ensuring trustworthy multimodal cognition in complex urban environments.
Evidence-Grounded Trustworthy Multimodal Reasoning and Evaluation Benchmark in Complex Urban Scenes
arXiv:2608.10954v2 Announce Type: replace-cross Abstract: While Multimodal Large Language Models (MLLMs) demonstrate impressive performance in benign scenarios, their cognitive reliability deteriorates significantly in complex scenes under adverse conditions. In these settings, models often rely on implicit inference without sufficient visual evidence, leading to a disconnect between perception and reasoning. Meanwhile, existing outcome-oriented benchmarks evaluate only final predictions and fail to diagnose failures in the underlying reasoning process. To address this gap, the authors propose AD2-Bench, which introduces a Hierarchical Visual Diagnosis framework that decomposes reasoning into a structured Chain of Evidence (CoE). This fine-grained diagnosis reveals that robust multimodal reasoning fundamentally depends on accurate evidence acquisition. Building on this perspective, the authors formulate reasoning from a probabilistic viewpoint and identify two primary causes of reasoning failure: Spatial Ambiguity, where models fail to distinguish target objects from background clutter, resulting in localization errors; and Semantic Uncertainty, where degraded visual features lead to incorrect semantic interpretation, resulting in understanding errors. To overcome these evidence deficiencies, they further propose Evidence-grounded Visual Reasoning (EGVOR), which replaces implicit reasoning with the explicit generation of Evidence Atoms - structured spatial-semantic triplets that enforce tight alignment between localization and semantic understanding. The model is trained through a hierarchical curriculum that progresses from reflective supervision construction to reinforcement learning, where reducing reasoning variance is explicitly rewarded. Extensive experiments demonstrate that EGVOR substantially improves reasoning stability under adverse conditions, providing a more robust framework for trustworthy multimodal cognition.
Abstract:While Multimodal Large Language Models (MLLMs) demonstrate impressive performance in benign scenarios, their cognitive reliability deteriorates significantly in complex scenes under adverse conditions. In these settings, models often rely on implicit inference without sufficient visual evidence, leading to a disconnect between perception and reasoning. Meanwhile, existing outcome-oriented benchmarks evaluate only final predictions and fail to diagnose failures in the underlying reasoning process. To address this gap, the authors propose AD2-Bench, which introduces a Hierarchical Visual Diagnosis framework that decomposes reasoning into a structured Chain of Evidence (CoE). This fine-grained diagnosis reveals that robust multimodal reasoning fundamentally depends on accurate evidence acquisition. Building on this perspective, the authors formulate reasoning from a probabilistic viewpoint and identify two primary causes of reasoning failure: Spatial Ambiguity, where models fail to distinguish target objects from background clutter, resulting in localization errors; and Semantic Uncertainty, where degraded visual features lead to incorrect semantic interpretation, resulting in understanding errors. To overcome these evidence deficiencies, they further propose Evidence-grounded Visual Reasoning (EGVOR), which replaces implicit reasoning with the explicit generation of Evidence Atoms - structured spatial-semantic triplets that enforce tight alignment between localization and semantic understanding. The model is trained through a hierarchical curriculum that progresses from reflective supervision construction to reinforcement learning, where reducing reasoning variance is explicitly rewarded. Extensive experiments demonstrate that EGVOR substantially improves reasoning stability under adverse conditions, providing a more robust framework for trustworthy multimodal cognition.
This submission is an iterative version of our previous work, **"AD^2-Bench: A Hierarchical CoT Benchmark for MLLM in Autonomous Driving under Adverse Conditions"** (arXiv:2506.09557). We plan to consolidate the current submission with the earlier version into a unified manuscript. Therefore, we would like to withdraw this submission
Computer Vision and Pattern Recognition (cs.CV); Artificial Intelligence (cs.AI)
arXiv:2608.10954 [cs.CV]
(or arXiv:2608.10954v2 [cs.CV] for this version)
Text integral sub licența deschisă a publicației · 298 cuvinte
Subiecte conexe
[AINews] Reality Checks on AI News (Yegge shuts down Gas Town, Databricks’ +60% Astra cost)
Latent Space · 17 sept. 2026
Evidence-Grounded Trustworthy Multimodal Reasoning and Evaluation Benchmark in Complex Urban Scenes
arXiv · 1 sept. 2026
Enabling a new model for healthcare with AI co-clinician
Google DeepMind · 30 apr. 2026
TopoGraphRAG-Bench: Evaluating Multimodal GraphRAG on Layout-Grounded Evidence Reasoning
arXiv · 8 oct. 2026