mlsanityData to 7 Oct 2026

Spotlight and Shadow: Attention-Guided Dual-Anchor Introspective Decoding for MLLM Hallucination Mitigation

arXiv 2604.10071 ↗First posted 11 Apr 2026cs.CVcited by 1

On the map: LLM reasoning → Hallucination detection and mitigation

also in 1 more
Abstract (arXiv)

Multimodal Large Language Models (MLLMs) have demonstrated remarkable reasoning capabilities yet continue to suffer from hallucination, where generated text contradicts visual content. In this paper, we introduce Dual-Anchor Introspective Decoding (DaID), a novel contrastive decoding framework that dynamically calibrates each token generation by mining the model's internal perceptual discrepancies. Specifically, DaID identifies a Spotlight layer to amplify visual factual signals and a Shadow layer to suppress textual inertia. By leveraging visual attention distributions to guide this dual-anchor selection process, our method ensures precise, token-specific adaptation. Experimental results across multiple benchmarks and MLLMs demonstrate that DaID significantly mitigates hallucination while enhancing general reasoning capabilities.

Key facts. Cited by 1 paper in the map (data through 7 Oct 2026). Placed in Hallucination detection and mitigation, one of 17 directions in LLM reasoning. Builds on 29 of its 46 references that are in the map. First posted on arXiv 11 Apr 2026.

Builds on

29 of 46 references.

Closest ancestors · 2

Same direction, newest first

  1. 2025Mitigating Object Hallucinations in Large Vision-Language Models via Attention Calibration4 Feb
  2. 2024Combating Multimodal LLM Hallucination via Bottom-Up Holistic Reasoning15 Dec

Other references · 27

  1. 2024Qwen2-VL: Enhancing Vision-Language Model's Perception of the World at Any Resolutioncited by 4,151
  2. 2023InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuningcited by 2,855
  3. 2016Making the V in VQA Matter: Elevating the Role of Image Understanding in Visual Question Answeringcited by 2,498
  4. 2023MMBench: Is Your Multi-modal Model an All-around Player?cited by 2,105
  5. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Modelscited by 1,624
  6. 2018VizWiz Grand Challenge: Answering Visual Questions from Blind Peoplecited by 852
  7. 2023Mitigating Object Hallucinations in Large Vision-Language Models through Visual Contrastive Decodingcited by 671
  8. 2018Object Hallucination in Image Captioningcited by 661
  9. 2022Contrastive Decoding: Open-ended Text Generation as Optimizationcited by 598
  10. 2023OPERA: Alleviating Hallucination in Multi-Modal Large Language Models via Over-Trust Penalty and Retrospection-Allocationcited by 468
  11. 2024Mitigating Hallucinations in Large Vision-Language Models with Instruction Contrastive Decodingcited by 198
  12. 2024MLLM can see? Dynamic Correction Decoding for Hallucination Mitigationcited by 87
15 more
  1. 2025Look-Back: Implicit Visual Re-focusing in MLLM Reasoningcited by 54
  2. 2024Personalizing Multimodal Large Language Models for Image Captioning: An Experimental Analysiscited by 23
  3. 2024Surveying the MLLM Landscape: A Meta-Review of Current Surveyscited by 20
  4. 2025Hallucinatory Image Tokens: A Training-free EAZY Approach on Detecting and Mitigating Object Hallucinations in LVLMscited by 14
  5. 2025VideoEraser: Concept Erasure in Text-to-Video Diffusion Modelscited by 12
  6. 2024Breaking the Memory Wall for Heterogeneous Federated Learning via Progressive Trainingcited by 10
  7. 2025Bridging the Copyright Gap: Do Large Vision-Language Models Recognize and Respect Copyrighted Content?cited by 8
  8. 2024MME-Finance: A Multimodal Finance Benchmark for Expert-level Understanding and Reasoningcited by 7
  9. 2024CopyrightMeter: Revisiting Copyright Protection in Text-to-image Modelscited by 6
  10. 2026TSEmbed: Unlocking Task Scaling in Universal Multimodal Embeddingscited by 6
  11. 2025Memory-Efficient Federated Fine-Tuning of Large Language Models via Layer Pruningcited by 5
  12. 2025Elastic Mixture of Rank-Wise Experts for Knowledge Reuse in Federated Fine-Tuningcited by 5
  13. 2026FraudShield: Knowledge Graph Empowered Defense for LLMs against Fraud Attackscited by 3
  14. 2026Beyond End-to-End: Dynamic Chain Optimization for Private LLM Adaptation on the Edgecited by 3
  15. 2026"I See What You Did There": Can Large Vision-Language Models Understand Multimodal Puns?cited by 1

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026StreamFlow: Dynamic Memory Flows for Streaming Video Understandingcited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025ASCD: Attention-Steerable Contrastive Decoding for Reducing Hallucination in MLLM
  2. 2026When Looking Is Not Enough: Visual Attention Structure Reveals Hallucination in MLLMs
  3. 2024VaLiD: Mitigating the Hallucination of Large Vision Language Models by Visual Layer Fusion Contrastive Decoding
  4. 2024Mitigating Hallucination in Visual-Language Models via Re-Balancing Contrastive Decoding
  5. 2026SAGE: Sink-Aware Grounded Decoding for Multimodal Hallucination Mitigation
  6. 2026Global Context or Local Detail? Adaptive Visual Grounding for Hallucination Mitigation
  7. 2025Attention Reallocation: Towards Zero-cost and Controllable Hallucination Mitigation of MLLMs
  8. 2026CHASD: Language Increment-Calibrated Contrastive Decoding against Hallucination in LVLMs