Multimodal Model Interpretability

Follow this topic Ask about this topic

It asks which modalities, regions, tokens, concepts, or internal circuits support a model’s output, helping people audit errors, shortcuts, bias, and safety. Common approaches include saliency and Shapley attribution, concept bottlenecks and sparse autoencoders, and probing or causal interventions that test what happens when evidence is altered. Researchers evaluate these explanations on visual question answering, hallucination and grounding tasks, cross-modal benchmarks, and clinical settings where inputs may be missing or incomplete. A central tension is that an explanation can look plausible without being faithful: models may encode useful evidence yet ignore it, favor one modality, or rely on shortcuts, and evaluation of genuine cross-modal reasoning remains difficult.

Latest papers

130 papers13 papers in the last four weeks, down 7% on the four weeks before. 0.1% of all new papers.

Physics-based phenomenological characterization of cross-modal bias in multimodal models
Feb 24, 2026Hyeongmo Kim, Sohyun Kang, Yerin Choi +5Multimodal Large Language Models Algorithmic Fairness
Brain Science Institute, Korea Institute of Science and Technology, Seoul, Republic of Korea · Department of Physics and Astronomy, Seoul National University, Seoul, Republic of Korea · Department of Computer Science and Engineering, Korea University, Seoul, Republic of Korea +2
DReX: An Explainable Deep Learning-based Multimodal Recommendation Framework
Feb 23, 2026Adamya Shyam, Venkateswara Rao Kagita, Bharti Rana +1Gated Recurrent Unit Multimodal Recommendation
University of Delhi, Delhi, India · National Institute of Technology, Warangal, India
FOCA: Frequency-Oriented Cross-Domain Forgery Detection, Localization and Explanation via Multi-Modal Large Language Model
Feb 21, 2026Zhou Liu, Tonghua Su, Hongshi Zhang +4Image Forgery Detection Digital Image Forensics
Harbin Institute of Technology, Harbin, China · DZ-Matrix · Guangdong Laboratory of Artificial Intelligence and Digital Economy (SZ), Shenzhen, China +2
Characterizing the Predictive Impact of Modalities with Supervised Latent-Variable Modeling
Feb 19, 2026Divyam Madaan, Sumit Chopra, Kyunghyun ChoMissing-Modality Learning Multimodal Learning
New York University. · New York University Grossman School of Medicine · Genentech +1
MRC-GAT: A Meta-Relational Copula-Based Graph Attention Network for Interpretable Multimodal Alzheimer's Disease Diagnosis
Feb 17, 2026Fatemeh Khalvandi, Saadat Izadi, Abdolah ChalechaleAlzheimer's Disease Classification Graph Attention Networks
Computer Engineering Department, Razi University, Kermanshah, Iran
Concept-Enhanced Multimodal RAG: Towards Interpretable and Accurate Radiology Report Generation
Feb 17, 2026Marco Salmè, Federico Siciliano, Fabrizio Silvestri +3Radiology Report Generation Radiology VLMs
Department of Engineering, Research Unit of Artificial Intelligence and Computer Systems, Università Campus Bio-Medico of Roma, Rome, Italy. · Department of Computer, Control and Management Engineering, Sapienza University of Rome, Rome, Italy. · Department of Diagnostics and Intervention, Radiation Physics, Biomedical Engineering, Umeå University, Umeå, Sweden. +1
Diagnosing Knowledge Conflict in Multimodal Long-Chain Reasoning
Feb 16, 2026Jing Tang, Kun Wang, Haolang Lu +7Knowledge Conflicts in Language Models Multimodal Large Language Models
Huazhong University of Science and Technology · Nanyang Technological University · Beijing University of Posts and Telecommunications +2
Attribution Graphs and Causal Probing for Mechanistic Discovery and Bias Repair in Multimodal Generative Learning
Oct 14, 2025Noor Islam S. Mohammad, Uluğ BayazıtGradient-Based Attribution Causal Representation Learning
1,2Istanbul Technical University
Evidence Recomposition and Predictive Context Residualization for Visual Attribution in Multimodal Large Language Models
Sep 26, 2025Jiawei Liang, Jianjie Huang, Ruoyu Chen +4Feature Attribution Multimodal Large Language Models
Shenzhen Campus of Sun Yat-sen University, Shenzhen 518107, China · Zhongguan-cun Academy, Beijing 100094, China · University of Chinese Academy of Sciences, Beijing 100049, China +2
Explainability in mulimodal deep transformation models for stroke outcome prediction
Apr 7, 2025Lisa Herzog, Jonas Brändli, Maurice Schneeberger +7Clinical Outcome Prediction Explainable Medical Image Analysis
University Hospital Zurich, Department of Neurology, Frauenklinikstrasse 26, 8091 Zurich, Switzerland · Zurich University of Applied Sciences, Institute of Data Analysis and Process Design (IDP), Technikumstrasse 9, 8401 Winterthur, Switzerland · University of Zurich, Epidemiology, Biostatistics and Prevention Institute (EBPI), Hirschengraben 84, 8001 Zurich, Switzerland +1

Feb 24, 2026J/K move · Enter open · S save

Physics-based phenomenological characterization of cross-modal bias in multimodal models

Hyeongmo Kim, Sohyun Kang, Yerin Choi, Seungyeon Ji+4

Brain Science Institute, Korea Institute of Science and Technology, Seoul, Republic of Korea · Department of Physics and Astronomy, Seoul National University, Seoul, Republic of Korea · Department of Computer Science and Engineering, Korea University, Seoul, Republic of Korea+2

Multimodal Large Language Models Algorithmic Fairness Multimodal Model Interpretability

Open paper AskPDF arXiv

The term 'algorithmic fairness' is used to evaluate whether AI models operate fairly in both comparative (where fairness is understood as formal equality, such as "treat like cases as like") and non-comparative (where unfairness arises from the model's inaccuracy, arbitrariness, or inscrutability) contexts. Recent advances in multimodal large language models (MLLMs) are breaking new ground in multimodal understanding, reasoning, and generation; however, we argue that inconspicuous distortions arising from complex multimodal interaction dynamics can lead to systematic bias. The purpose of this position paper is twofold: first, it is intended to acquaint AI researchers with phenomenological explainable approaches that rely on the physical entities that the machine experiences during training/inference, as opposed to the traditional cognitivist symbolic account or metaphysical approaches; second, it is to state that this phenomenological doctrine will be practically useful for tackling algorithmic fairness issues in MLLMs. We develop a surrogate physics-based model that describes transformer dynamics (i.e., semantic network structure and self-/cross-attention) to analyze the dynamics of cross-modal bias in MLLM, which are not fully captured by conventional embedding- or representation-level analyses. We support this position through multi-input diagnostic experiments: 1) perturbation-based analyses of emotion classification using Qwen2.5-Omni and Gemma 3n, and 2) dynamical analysis of Lorenz chaotic time-series prediction through the physical surrogate. Across two architecturally distinct MLLMs, we show that multimodal inputs can reinforce modality dominance rather than mitigate it, as revealed by structured error-attractor patterns under systematic label perturbation, complemented by dynamical analysis.