mlsanityData to 14 Oct 2026

A Survey on Safe Multi-Modal Learning System

arXiv 2402.05355 ↗First posted 8 Feb 2024cs.CY cs.AIcited by 15

Not on the map. Why?

Abstract (arXiv)

In the rapidly evolving landscape of artificial intelligence, multimodal learning systems (MMLS) have gained traction for their ability to process and integrate information from diverse modality inputs. Their expanding use in vital sectors such as healthcare has made safety assurance a critical concern. However, the absence of systematic research into their safety is a significant barrier to progress in this field. To bridge the gap, we present the first taxonomy that systematically categorizes and assesses MMLS safety. This taxonomy is structured around four fundamental pillars that are critical to ensuring the safety of MMLS: robustness, alignment, monitoring, and controllability. Leveraging this taxonomy, we review existing methodologies, benchmarks, and the current state of research, while also pinpointing the principal limitations and gaps in knowledge. Finally, we discuss unique challenges in MMLS safety. In illuminating these challenges, we aim to pave the way for future research, proposing potential directions that could lead to significant advancements in the safety protocols of MMLS.

Key facts. Cited by 15 papers in the map (data through 14 Oct 2026). Builds on 72 of its 112 references that are in the map. First posted on arXiv 8 Feb 2024.

Builds on

72 of 112 references.

References · 72

  1. 2022Training language models to follow instructions with human feedbackcited by 15,971
  2. 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 8,065
  3. 2017On Calibration of Modern Neural Networkscited by 4,960
  4. 2020Learning to summarize from human feedbackcited by 2,588
  5. 2016V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentationcited by 2,292
  6. 2019A Survey on Bias and Fairness in Machine Learningcited by 1,917
  7. 2022Model soups: averaging weights of multiple fine-tuned models improves accuracy without increasing inference timecited by 1,396
  8. 2017Multimodal Machine Learning: A Survey and Taxonomycited by 1,013
  9. 2023Aligning Large Multimodal Models with Factually Augmented RLHFcited by 712
  10. 2018Real-world Anomaly Detection in Surveillance Videoscited by 577
  11. 2017Future Frame Prediction for Anomaly Detection -- A New Baselinecited by 391
  12. 2023Visual Adversarial Examples Jailbreak Aligned Large Language Modelscited by 328
38 more
  1. 2020Large-Scale Adversarial Training for Vision-and-Language Representation Learningcited by 327
  2. 2023WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentationcited by 315
  3. 2022A Survey of Machine Unlearningcited by 312
  4. 2023Forget-Me-Not: Learning to Forget in Text-to-Image Diffusion Modelscited by 301
  5. 2023AnomalyCLIP: Object-agnostic Prompt Learning for Zero-shot Anomaly Detectioncited by 271
  6. 2019On Completeness-aware Concept-Based Explanations in Deep Neural Networkscited by 239
  7. 2020Not only Look, but also Listen: Learning Multimodal Violence Detection under Weak Supervisioncited by 238
  8. 2021VT-ADL: A Vision Transformer Network for Image Anomaly Detection and Localizationcited by 223
  9. 2023Understanding and Mitigating Copying in Diffusion Modelscited by 209
  10. 2022Are Multimodal Transformers Robust to Missing Modality?cited by 186
  11. 2022Domino: Discovering Systematic Errors with Cross-Modal Embeddingscited by 156
  12. 2023On the Adversarial Robustness of Multi-Modal Foundation Modelscited by 145
  13. 2022Towards Adversarial Attack on Vision-Language Pre-training Modelscited by 144
  14. 2023Multimodal Industrial Anomaly Detection via Hybrid Fusioncited by 139
  15. 2022Data Determines Distributional Robustness in Contrastive Language Image Pre-training (CLIP)cited by 138
  16. 2023Silkie: Preference Distillation for Large Visual Language Modelscited by 130
  17. 2020Improving model calibration with accuracy versus uncertainty optimizationcited by 123
  18. 2023VadCLIP: Adapting Vision-Language Models for Weakly Supervised Video Anomaly Detectioncited by 123
  19. 2023AdvCLIP: Downstream-agnostic Adversarial Examples in Multimodal Contrastive Learningcited by 103
  20. 2021Mixed supervision for surface-defect detection: from weakly to fully supervised learningcited by 101
  21. 2023Multimodal Federated Learning via Contrastive Representation Ensemblecited by 86
  22. 2023Provable Dynamic Fusion for Low-Quality Multimodal Datacited by 82
  23. 2024Red Teaming Visual Language Modelscited by 65
  24. 2021The Spotlight: A General Method for Discovering Systematic Errors in Deep Learning Modelscited by 63
  25. 2023CLIP-AD: A Language-Guided Staged Dual-Path Model for Zero-shot Anomaly Detectioncited by 58
  26. 2021Socially Responsible AI Algorithms: Issues, Purposes, and Challengescited by 56
  27. 2022Robust Cross-Modal Representation Learning with Progressive Self-Distillationcited by 55
  28. 2023FedMultimodal: A Benchmark For Multimodal Federated Learningcited by 51
  29. 2018VQA-E: Explaining, Elaborating, and Enhancing Your Answers for Visual Questionscited by 50
  30. 2021A Review on Explainability in Multimodal Deep Neural Netscited by 46
  31. 2023Can Language Models be Instructed to Protect Personal Information?cited by 46
  32. 2023A Content-Driven Micro-Video Recommendation Dataset at Scalecited by 45
  33. 2023Myriad: Large Multimodal Model by Applying Vision Experts for Industrial Anomaly Detectioncited by 39
  34. 2023Practical Membership Inference Attacks Against Large-Scale Multi-Modal Models: A Pilot Studycited by 38
  35. 2020A Closer Look at the Robustness of Vision-and-Language Pre-trained Modelscited by 37
  36. 2019Well-calibrated Model Uncertainty with Temperature Scaling for Dropout Variational Inferencecited by 36
  37. 2021Assessing Multilingual Fairness in Pre-trained Multimodal Representationscited by 35
  38. 2023Towards Calibrated Robust Fine-Tuning of Vision-Language Modelscited by 35

50 most cited of 72 All 72 (JSON)

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2024: 3; 2025: 9; 2026: 3.

15 papers.

Sort:

By area:

  1. 2025A Survey on Mechanistic Interpretability for Multi-Modal Foundation Modelscited by 34
  2. 2025Large Vision-Language Model Alignment and Misalignment: A Survey Through the Lens of Explainabilitycited by 28
  3. 2025MathAgent: Leveraging a Mixture-of-Math-Agent Framework for Real-World Multimodal Mathematical Error Detectioncited by 24
  4. 2024Surveying the MLLM Landscape: A Meta-Review of Current Surveyscited by 20
  5. 2024The Synergy between Data and Multi-Modal Large Language Models: A Survey from Co-Development Perspectivecited by 15
  6. 2025The Eye of Sherlock Holmes: Uncovering User Private Attribute Profiling via Vision-Language Model Agentic Frameworkcited by 9
  7. 2025Multimodal Cultural Safety: Evaluation Framework and Alignment Strategiescited by 8
  8. 2025Synergistic Prompting for Robust Visual Recognition with Missing Modalitiescited by 8
  9. 2025Improving Generalizability and Undetectability for Targeted Adversarial Attacks on Multimodal Pre-trained Modelscited by 5
  10. 2025Towards deployment-centric multimodal AI beyond vision and languagecited by 3
  11. 2024How Private are Language Models in Abstractive Summarization?cited by 2
  12. 2026Lingua-SafetyBench: A Benchmark for Safety Evaluation of Multilingual Vision-Language Modelscited by 2
3 more
  1. 2026An Empirical Analysis of Calibration and Selective Prediction in Multimodal Clinical Condition Classificationcited by 2
  2. 2025Lyapunov-Stable Adaptive Control for Multimodal Concept Driftcited by 0
  3. 2026MAPLE: Modality-Aware Post-training and Learning Ecosystemcited by 0