mlsanityData to 14 Oct 2026

Concurrent Image Understanding and Generation: Self-Correcting Coupled Markov Jump Processes

arXiv 2607.13188 ↗First posted 14 Jul 2026cs.LGcited by 1

Not on the map. Why?

Abstract (arXiv)

Human cognition does not separate understanding and generation. A teacher at a whiteboard speaks and draws textittogether, each modality reshapes the other. In this paper, we bring this coupled loop to artificial systems. Masked Diffusion Models (MDMs) are ideally suited to this task, yet existing samplers either decode text and image interleavedly or independently update them in parallel branches that share only previous-step history, but not the other modality's latest decisions textitwithin the same step; combined with MDMs' inability to remask, cross-modal contradictions are neither detected nor repaired. We introduce Self-Correcting Coupled Markov Jump Processes (SC-CMJP), a framework in which one modality's transition rates are functionals of the other modality's confidence score, as weighted by cross-modal attention. Furthermore, a remasking jump retracts commitments the moment cross-modal evidence turns against them. In conjunction with SC-CMJP, we introduce textttCO_texttt2textttJump (Self-underlineCOrrecting underlineCOupled underlineJump), a novel training-free single-pass sampler for joint multimodal geneneration. For training and evaluation purposes, we have created and will release three large-scale joint multimodal generation corpora: JEdit-1M, JMaze-200K, JNono-200K, with matching in- and out-of-distribution benchmarks. textttCO_texttt2textttJump achieves best joint performance for image understanding and editing as well as visual reasoning (maze and nonogram solving). The performance of the sampler scales monotonically with the number of denoising steps, evidence that the benefits of cross-modal coupling textitcompound across the trajectory. Project page: https://coupled-jump.github.io

Key facts. Cited by 1 paper in the map (data through 14 Oct 2026). Builds on 32 of its 39 references that are in the map. First posted on arXiv 14 Jul 2026.

Builds on

32 of 39 references.

References · 32

  1. 2014Microsoft COCO: Common Objects in Contextcited by 17,608
  2. 2017Neural Discrete Representation Learningcited by 4,712
  3. 2021Structured Denoising Diffusion Models in Discrete State-Spacescited by 1,719
  4. 2025Large Language Diffusion Modelscited by 912
  5. 2025Emerging Properties in Unified Multimodal Pretrainingcited by 897
  6. 2024Simple and Effective Masked Diffusion Language Modelscited by 848
  7. 2025Janus-Pro: Unified Multimodal Understanding and Generation with Data and Model Scalingcited by 812
  8. 2023Discrete Diffusion Modeling by Estimating the Ratios of the Data Distributioncited by 745
  9. 2024Simplified and Generalized Masked Diffusion for Discrete Datacited by 626
  10. 2025Dream 7B: Diffusion Large Language Modelscited by 544
  11. 2022A Continuous Time Framework for Discrete Denoising Modelscited by 451
  12. 2024Your Absorbing Discrete Diffusion Secretly Models the Conditional Distributions of Clean Datacited by 381
20 more
  1. 2024Discrete Flow Matchingcited by 317
  2. 2024Generative Flows on Discrete State-Spaces: Enabling Multimodal Flows with Applications to Protein Co-Designcited by 305
  3. 2025MMaDA: Multimodal Large Diffusion Language Modelscited by 250
  4. 2025Show-o2: Improved Native Unified Multimodal Modelscited by 250
  5. 2025ImgEdit: A Unified Image Editing Dataset and Benchmarkcited by 231
  6. 2025Train for the Worst, Plan for the Best: Understanding Token Ordering in Masked Diffusionscited by 214
  7. 2025Remasking Discrete Diffusion Models with Inference-Time Scalingcited by 189
  8. 2024OmniEdit: Building Image Editing Generalist Models Through Specialist Supervisioncited by 145
  9. 2025SDAR: A Synergistic Diffusion-AutoRegression Paradigm for Scalable Sequence Generationcited by 129
  10. 2025Lumina-DiMOO: An Omni Diffusion Large Language Model for Multi-Modal Generation and Understandingcited by 88
  11. 2025Path Planning for Masked Diffusion Model Samplingcited by 80
  12. 2025Generalized Interpolating Discrete Diffusioncited by 70
  13. 2024Informed Correctors for Discrete Diffusion Modelscited by 57
  14. 2025Fine-Tuning Masked Diffusion for Provable Self-Correctioncited by 39
  15. 2025Unified Diffusion VLA: Vision-Language-Action Model via Joint Discrete Denoising Diffusion Processcited by 39
  16. 2025Don't Settle Too Early: Self-Reflective Remasking for Diffusion Language Modelscited by 37
  17. 2023A Configurable Library for Generating and Manipulating Maze Datasetscited by 31
  18. 2024Foundation Inference Models for Markov Jump Processescited by 13
  19. 2025MMaDA-Parallel: Multimodal Large Diffusion Language Models for Thinking-Aware Editing and Generationcited by 8
  20. 2026Training-Free Self-Correction for Multimodal Masked Diffusion Modelscited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 1.

1 paper.

  1. 2026VBVR-Pro: A Scalable and Verifiable Suite for Native Visual Reasoningcited by 6

Similar, not cited

Close in wording, no citation either way

  1. 2025Diffusion Counterfactual Generation with Semantic Abduction
  2. 2025Multi-modal Synthetic Data Training and Model Collapse: Insights from VLMs and Diffusion Models
  3. 2026The Design Space of Tri-Modal Masked Diffusion Models
  4. 2025From Denoising to Refining: A Corrective Framework for Vision-Language Diffusion Model
  5. 2024Dual Diffusion for Unified Image Generation and Understanding
  6. 2026PreDiff-LM: Pretrained Discrete Masked Diffusion Language Modeling with Hybrid Attention
  7. 2026Images Talk Back: Guiding MMDiTs by Suppressing Image-to-Text Feedback
  8. 2026Coupled Inference in Diffusion Models for Semantic Decomposition