Skip to content
AI News HubLIVE
Original source2 min read

Step Back to Move Forward: Reflection-Aware Preference Optimization for Visual Generation

Summary

This paper introduces RA-GRPO, a reinforcement-learning-based preference alignment framework for diffusion visual generation. It incorporates backward reflection into forward generation by correcting intermediate sampling trajectories via diffusion inversion and distilling them with counterfactual path synthesis, reducing reward hacking and improving generalization in text-to-image and text-to-video models without architectural changes or inference overhead.

SourcearXiv Computer VisionAuthor: Junlong Wu, Jiuzhou Lin, Jia Sun, Boheng Zhang, Huaiqing Wang, Dewen Fan, Houde Liu, Qianqian Gan, Fan Yang, Tingting Gao
Step Back to Move Forward: Reflection-Aware Preference Optimization for Visual Generation
Report an error

The correction channel is not available yet. You can copy the article reference below for later.

Correction instructions
Read article

[Submitted on 3 Sep 2026]

Title:Step Back to Move Forward: Reflection-Aware Preference Optimization for Visual Generation

View a PDF of the paper titled Step Back to Move Forward: Reflection-Aware Preference Optimization for Visual Generation, by Junlong Wu and 9 other authors

View PDF HTML (experimental)

Abstract:Diffusion models have become the mainstream paradigm for modern visual generation and have substantially advanced multimedia content synthesis, especially in text-to-image and text-to-video tasks. To further align such generative models with human preferences, reinforcement learning (RL) has recently shown strong potential as a post-training strategy. Nevertheless, existing policy gradient-based methods often explore inefficiently, making them vulnerable to local optima that may degrade semantic faithfulness and visual realism. To address these challenges, we present Reflection-Aware GRPO (RA-GRPO), a new RL-based preference alignment framework for diffusion generative models. The core idea is to improve "forward" generation by incorporating "backward" reflection during optimization. We first introduce Diffusion Reflection, which rectifies intermediate sampling trajectories by inverting the diffusion process with a weak estimator, guiding latent states toward higher-probability regions of the true data manifold. Furthermore, we introduce Counterfactual Path Synthesis to implicitly distill these rectified trajectories into the policy, enabling the model to internalize the benefits of search-based exploration without incurring inference-time overhead. Extensive experiments on T2I and T2V models demonstrate that RA-GRPO significantly outperforms existing methods, particularly in mitigating reward hacking and improving generalization. The method remains architecture-agnostic and integrates seamlessly with standard pipelines, suggesting a promising direction for stable preference alignment.

Subjects:

Computer Vision and Pattern Recognition (cs.CV)

Cite as: arXiv:2609.04282 [cs.CV]

(or arXiv:2609.04282v1 [cs.CV] for this version)

https://doi.org/10.48550/arXiv.2609.04282

arXiv-issued DOI via DataCite (pending registration)

Submission history

From: Junlong Wu [view email] [v1] Thu, 3 Sep 2026 04:37:19 UTC (4,602 KB)

Full-text links:

Access Paper:

View a PDF of the paper titled Step Back to Move Forward: Reflection-Aware Preference Optimization for Visual Generation, by Junlong Wu and 9 other authors

View PDF

HTML (experimental)

TeX Source

view license

Current browse context:

cs.CV

new | recent | 2026-09

Change to browse by:

cs

References & Citations

NASA ADS

Google Scholar

Semantic Scholar

Loading...

Data provided by:

Bibliographic Tools

Bibliographic and Citation Tools

Bibliographic Explorer Toggle

Bibliographic Explorer (What is the Explorer?)

Connected Papers Toggle

Connected Papers (What is Connected Papers?)

Litmaps Toggle

Litmaps (What is Litmaps?)

scite.ai Toggle

scite Smart Citations (What are Smart Citations?)

Code, Data, Media

Code, Data and Media Associated with this Article

alphaXiv Toggle

alphaXiv (What is alphaXiv?)

Links to Code Toggle

CatalyzeX Code Finder for Papers (What is CatalyzeX?)

DagsHub Toggle

DagsHub (What is DagsHub?)

GotitPub Toggle

Gotit.pub (What is GotitPub?)

Huggingface Toggle

Hugging Face (What is Huggingface?)

ScienceCast Toggle

ScienceCast (What is ScienceCast?)

Demos

Demos

Replicate Toggle

Replicate (What is Replicate?)

Spaces Toggle

Hugging Face Spaces (What is Spaces?)

Spaces Toggle

TXYZ.AI (What is TXYZ.AI?)

Related Papers

Recommenders and Search Tools

Link to Influence Flower

Influence Flower (What are Influence Flowers?)

Core recommender toggle

CORE Recommender (What is CORE?)

Author

Venue

Institution

Topic

About arXivLabs

arXivLabs: experimental projects with community collaborators

arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.

Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.

Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.

Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)

Key points and analysis

Article intelligence

EngineersAdvanced

Key points

  • Addresses inefficient exploration and local optima in policy-gradient RL fine-tuning of diffusion models.
  • Diffusion Reflection inverts the diffusion process with a weak estimator to guide latent states toward high-probability regions of the true data manifold.
  • Counterfactual Path Synthesis implicitly distills rectified trajectories into the policy, avoiding inference-time search overhead.
  • Achieves strong results on T2I and T2V models, mitigating reward hacking, improving generalization, and integrating with standard pipelines.

Highlights and analysis are generated automatically and may contain errors. Check the original source.

More recent stories on this topic · Models