Pith. sign in

REVIEW 3 major objections 5 minor 61 references

Diffusion classifiers fail on rare concepts because the model cannot generate them well; training it to prefer minority samples fixes the recognition gap without new images.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

T0 review · grok-4.5

2026-07-12 00:03 UTC pith:QOJ7JEE3

load-bearing objection Practical self-improvement loop that turns a reconstruction-error minority reward into modest diffusion-classifier gains without extra images; the mechanism claim is under-validated but the empirical package is real enough to engage. the 3 major comments →

arxiv 2607.03770 v1 pith:QOJ7JEE3 submitted 2026-07-04 cs.CV cs.AIcs.LG

Self-Improving Diffusion Classifiers with Minority Preference Optimization

Hyunsoo Kim , Jungmyung Wi , Soobin Um , Donghyun Kim , Suhyun Kim This is my paper
classification cs.CV cs.AIcs.LG
keywords diffusion classifierminority samplingpreference optimizationzero-shot classificationLoRAGRPOtext-to-image diffusion
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Diffusion models used as zero-shot classifiers work well on common, high-density visual concepts but struggle on rare ones. The paper argues this is not just a classification quirk: recognition tracks generative coverage of the data manifold, so what the model cannot create it cannot recognize. The authors introduce MiPO, a self-improving fine-tune that uses only arbitrary text captions. It generates candidate images, scores them with a minority preference reward based on reconstruction discrepancy, and updates a small LoRA adapter with group-relative policy optimization plus KL regularization. The result is prompt-adaptive minority generation that broadens low-density coverage and raises zero-shot classification accuracy on standard and out-of-distribution datasets, especially for minority groups, without extra images or external reward models.

Core claim

The perception ability of a diffusion classifier is biased toward majority regions of the pretrained data manifold. Expanding generative coverage of minority regions through minority preference optimization directly improves zero-shot recognition of underrepresented concepts, and this can be done from captions alone by self-generating samples and rewarding those the model reconstructs poorly.

What carries the argument

Minority Preference Optimization (MiPO): a LoRA-adapted Group Relative Policy Optimization loop whose reward is the DDIM reconstruction discrepancy M(x0)=||x̂0−x0||2, used as a proxy for how minority-like a sample is, with updates restricted to early denoising steps and KL regularization to the pretrained policy.

Load-bearing premise

The method treats higher DDIM reconstruction error at a fixed intermediate noise level as a faithful marker of true low-density minority regions rather than mere reconstruction difficulty or noise sensitivity.

What would settle it

Define minority and majority groups on a held-out set with an independent density estimator (for example kNN in a frozen feature space), apply the trained MiPO LoRA, and check whether the reported accuracy gains remain; if they vanish or reverse under the independent definition, the reconstruction proxy is not measuring true minority coverage.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Zero-shot diffusion classifiers can be strengthened without collecting additional images or external preference models.
  • Gains in generative coverage of low-density regions translate into better recognition of long-tail visual categories.
  • A compact LoRA adapter enables prompt-adaptive minority generation that can be attached or detached at inference with negligible overhead.
  • Methods that only guide minority sampling at test time without updating the model leave the classifier's internal bias largely unchanged.
  • The same self-improving loop works across different Stable Diffusion backbones and multiple classification benchmarks.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Because the minority signal is taken from the model's own reconstruction error, the loop could reinforce the model's blind spots; an independent density estimator would test whether true manifold coverage is improving.
  • The same coverage-to-recognition link suggests a reverse loop: use classification failures as free labels for further minority preference training.
  • A reconstruction-error minority proxy may transfer to other generative families that admit a Tweedie-style posterior mean, such as flow or consistency models.
  • Plug-and-play minority LoRAs could become a lightweight fairness module for generative models that must handle rare visual concepts without full retraining.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. The paper argues that diffusion classifiers are biased toward high-density (majority) regions of the pretrained data manifold and that this bias can be reduced by strengthening minority sampling. It introduces MiPO: a LoRA-based GRPO fine-tuning procedure that uses only arbitrary captions, samples multiple SDE trajectories per prompt, and rewards samples by a DDIM reconstruction discrepancy M(x0)=||x̂0−x0||2 at an intermediate noise level (t≈0.9T), with KL regularization and selective early-timestep updates. The resulting adapter is claimed to expand low-density generative coverage and thereby improve zero-shot diffusion classification on CIFAR-10, CIFAR-10-C, ImageNet-Tiny, Caltech, and SUN09 (gains up to ~3.8 points on SD 1.5/2.0) without extra images or external reward models, while remaining competitive on minority-oriented generation metrics.

Significance. If the claimed mechanism holds, the work would be a useful contribution: it links minority sampling to diffusion-classifier perception, shows that a compact LoRA+GRPO adapter can improve zero-shot classification without images or external rewards, and provides modular, prompt-adaptive minority generation. Strengths include the self-improving setup (captions only), ablations isolating KL and early-timestep selection (Tables 2–3), dual evaluation on classification and minority generation (Tables 1, 4–5), and practical plug-and-play design. The result would matter for generative classifiers and long-tail robustness, provided the reward is shown to target true low-density coverage rather than reconstruction hardness alone.

major comments (3)
  1. Sec. 3.2, Eqs. (3)–(4): The central mechanism rests on M(x0)=||x̂0−x0||2 (DDIM reconstruction discrepancy at fixed t≈0.9T) as a faithful proxy for low-density minority regions. M is computed from the same pretrained denoiser being fine-tuned, so it primarily measures reconstruction uncertainty under that model. The manuscript never validates that high-M samples lie in low-density regions of the true data distribution (or of an independent density estimator). The KNN density split is used only for post-hoc classifier evaluation (Sec. 5.1), not for reward validation. Without such a check (e.g., correlation of M with held-out density, or comparison to an external density model), the accuracy gains in Table 1 could arise from side-effects of GRPO+LoRA rather than the claimed manifold-coverage mechanism.
  2. Table 1 vs. Table 6: Gains reverse on LabelME and VOC2007 (drops of several points for both SD 1.5 and SD 2.0). The paper attributes this to multi-object/noisy annotations and defers analysis to the Appendix, but these are standard VLCS-style benchmarks. The claim that minority preference “broadens coverage … thereby improving diffusion-based recognition” needs either (i) a clear characterization of when the method helps vs. hurts, or (ii) evidence that the reverse results are not simply distributional shift away from the pretrained prior. As written, the mixed outcomes weaken the generality of the central claim.
  3. Tables 1–3 and experimental protocol: No error bars, multiple seeds, or statistical tests are reported, despite stochastic SDE sampling, GRPO, and random 2,000-image subsets for CIFAR-10-C and ImageNet-Tiny. Several reported lifts are modest (e.g., ~0.8–2 points). Without variance estimates it is hard to judge whether the improvements are reliable, especially given the free parameters α, β, early-timestep fraction, and perturbation timestep. At minimum, multi-seed means and standard deviations (or bootstrap intervals) on the main tables are needed to support the quantitative claims.
minor comments (5)
  1. Fig. 1 and abstract claim “up to 3.8%” gains; Table 1 shows smaller lifts on several sets. Align the headline number with the table or clarify which setting produces 3.8%.
  2. Sec. 3.2: α=0.7, β=0.15 and the 40% early-timestep fraction are stated without sensitivity analysis beyond the binary ablations in Tables 2–3. A short sensitivity plot or grid would strengthen reproducibility.
  3. Table 4: “JEPA-SCORE ↓” and Minority Score are useful, but the paper should briefly define how JEPA-SCORE is computed and why lower is better for minority coverage.
  4. Notation: πθ(at|xt) vs. pθ(xt−1|xt,c) and z vs. x are used somewhat interchangeably; a short consistency pass would help.
  5. Related work: briefly contrast with other reconstruction-error or uncertainty-based density proxies so the novelty of the training-time use of M is clearer.

Circularity Check

1 steps flagged

Mild self-referentiality in the minority reward (reconstruction error of the same denoiser family) but no by-construction reduction of the claimed accuracy gains; evaluations use independent metrics.

specific steps
  1. self definitional [Sec. 3.2, Eqs. (3)–(4) and surrounding text]
    "The minority score is then defined as the reconstruction discrepancy between the generated sample and its DDIM-approximated posterior mean: M(x0)=∥x̂0−x0∥2. Intuitively, samples with higher reconstruction errors indicate regions where the denoiser exhibits greater uncertainty or weaker representation, corresponding to minority areas of the data manifold."

    Minority preference is defined directly as the (pre)trained denoiser’s own reconstruction error at fixed t=0.9T. Preferring high-M samples is therefore, by construction, preferring samples the current model family reconstructs poorly; the subsequent claim that this equals broadened coverage of true low-density manifold regions rests on this definitional identification rather than an independent density derivation.

full rationale

The paper is an empirical methods contribution: it defines a reward from DDIM reconstruction discrepancy of a (pre)trained denoiser, optimizes a LoRA policy via GRPO+KL, and reports external zero-shot classification accuracies plus independent KNN density splits. There is no mathematical derivation chain in which a claimed prediction or first-principles result is algebraically forced by its inputs. The only mild circularity is definitional identification of 'minority' with high reconstruction error under the model being improved; this is an assumption/intuition (explicitly labeled as such) rather than a load-bearing uniqueness theorem or fitted-parameter-as-prediction. Self-citations to prior Um et al. minority-guidance papers supply the proxy idea but are not used to forbid alternatives or force the classifier gains. Classification and KNN results remain external and falsifiable, so overall circularity is low.

Axiom & Free-Parameter Ledger

4 free parameters · 3 axioms · 1 invented entities

The central claim rests on a small set of free hyper-parameters, standard diffusion/RL assumptions, and one paper-specific proxy (the reconstruction-error minority score) that is not independently validated outside the training loop.

free parameters (4)
  • α (minority reward weight) = 0.7
    Set to 0.7 by hand to balance reward vs. KL; no sensitivity sweep reported beyond the binary ablation.
  • β (KL coefficient) = 0.15
    Set to 0.15; controls how far the policy may deviate from the pretrained prior.
  • early-timestep fraction = 0.4
    Only the first 40 % of denoising steps are updated; chosen empirically.
  • perturbation timestep for minority score = 0.9T
    Fixed at t=0.9T for the DDIM reconstruction error used as reward.
axioms (3)
  • domain assumption Noise-prediction error of a diffusion model is a valid surrogate for class-conditional likelihood (standard diffusion-classifier assumption).
    Inherited from Li et al. / Clark & Jaini; used throughout Sec. 3.1 and evaluation.
  • ad hoc to paper Higher DDIM reconstruction discrepancy at intermediate noise indicates a sample lies in a low-density minority region.
    Core of the minority preference reward (Eqs. 3–4); not proven, only motivated by prior guidance literature.
  • domain assumption Group-relative advantages computed from SDE trajectories yield stable policy gradients for diffusion models.
    Taken from DanceGRPO / FlowGRPO literature and applied without re-derivation.
invented entities (1)
  • Minority Preference Reward M(x0) no independent evidence
    purpose: Provides a scalar training signal that prefers samples the model itself reconstructs poorly, thereby encouraging coverage of underrepresented regions.
    Defined solely inside the paper as ||x̂0-x0||2; no external density ground-truth or theoretical guarantee is supplied.

pith-pipeline@v1.1.0-grok45 · 19173 in / 2603 out tokens · 18763 ms · 2026-07-12T00:03:29.916937+00:00 · methodology

0 comments
read the original abstract

Prior studies have demonstrated that diffusion classifiers achieve robust zero-shot classification performance. However, their effectiveness is strongly tied to the pretraining data distribution: they perform well in majority, high-density regions of the data manifold, but are significantly less accurate in minority, low-density regions. Although prior works on minority sampling have focused on generating more minority-like images, what minority sampling fundamentally enables beyond generation remains underexplored. In this paper, we reveal a direct relationship between minority sampling in generation and the perception capability of diffusion classifiers. Specifically, we show that enhancing minority sampling broadens the coverage of underrepresented regions on the data manifold, thereby improving diffusion-based recognition. To exploit this connection, we propose \textit{Self-Improving Diffusion Classifiers with Minority Preference Optimization} (MiPO), which fine-tunes a pretrained diffusion model using minority preference rewards. Using only arbitrary caption data, MiPO generates candidate samples, rewards those that better cover minority regions, and optimizes the model with LoRA and Group Relative Policy Optimization, without additional image data, external foundation models, or external reward models. This enables stable, prompt-adaptive minority sampling and translates low-density generative coverage into improved zero-shot diffusion classification. To sum up, we show that diffusion classifier perception is biased toward majority regions, demonstrate that this bias can be alleviated through minority preference optimization, and evaluate MiPO on five standard datasets.

Figures

Figures reproduced from arXiv: 2607.03770 by Donghyun Kim, Hyunsoo Kim, Jungmyung Wi, Soobin Um, Suhyun Kim.

Figure 1
Figure 1. Figure 1: MiPO expands the perceptual coverage of diffusion classifiers. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Training pipeline of Minority Preference Optimization (MiPO). [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Qualitative results of minority samples. [PITH_FULL_IMAGE:figures/full_fig_p012_3.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

61 extracted references · 19 linked inside Pith

  1. [1]

    arXiv preprint arXiv:1409.0473 (2014)

    Bahdanau, D.: Neural machine translation by jointly learning to align and trans- late. arXiv preprint arXiv:1409.0473 (2014)

    Pith/arXiv arXiv 2014

  2. [2]

    arXiv preprint arXiv:2510.05949 (2025)

    Balestriero, R., Ballas, N., Rabbat, M., LeCun, Y.: Gaussian embeddings: How jepas secretly learn your data density. arXiv preprint arXiv:2510.05949 (2025)

    arXiv 2025

  3. [3]

    arXiv preprint arXiv:2305.13301 (2023)

    Black, K., Janner, M., Du, Y., Kostrikov, I., Levine, S.: Training diffusion models with reinforcement learning. arXiv preprint arXiv:2305.13301 (2023)

    Pith/arXiv arXiv 2023

  4. [4]

    arXiv preprint arXiv:2505.18963 (2025)

    Chan-Santiago, J.A., Tirupattur, P., Nayak, G.K., Liu, G., Shah, M.: MGD3: Mode-guided dataset distillation using diffusion models. arXiv preprint arXiv:2505.18963 (2025)

    Pith/arXiv arXiv 2025

  5. [5]

    arXiv preprint arXiv:2305.15241 (2023)

    Chen, H., Dong, Y., Wang, Z., Yang, X., Duan, C., Su, H., Zhu, J.: Robust classi- fication via a single diffusion model. arXiv preprint arXiv:2305.15241 (2023)

    Pith/arXiv arXiv 2023

  6. [6]

    URL https://arxiv

    Chen, W., Zhang, J., Wu, J., Wu, H., Xiao, X., Lin, L.: Id-aligner: Enhancing identity-preserving text-to-image generation with reward feedback learning, 2024. URL https://arxiv. org/abs/2404.15449

    Pith/arXiv arXiv 2024

  7. [7]

    arXiv preprint arXiv:2510.12060 (2025)

    Chen, Y.C., Inouye, D.I., Gao, J.: Your var model is secretly an efficient and explainable generative classifier. arXiv preprint arXiv:2510.12060 (2025)

    arXiv 2025

  8. [8]

    In: 2010 IEEE computer society conference on computer vision and pattern recognition

    Choi, M.J., Lim, J.J., Torralba, A., Willsky, A.S.: Exploiting hierarchical context on a large database of object categories. In: 2010 IEEE computer society conference on computer vision and pattern recognition. pp. 129–136. IEEE (2010)

    2010

  9. [9]

    arXiv preprint arXiv:2209.14687 (2022)

    Chung, H., Kim, J., Mccann, M.T., Klasky, M.L., Ye, J.C.: Diffusion posterior sam- pling for general noisy inverse problems. arXiv preprint arXiv:2209.14687 (2022)

    Pith/arXiv arXiv 2022

  10. [10]

    Ad- vances in Neural Information Processing Systems36, 58921–58937 (2023)

    Clark, K., Jaini, P.: Text-to-image diffusion models are zero shot classifiers. Ad- vances in Neural Information Processing Systems36, 58921–58937 (2023)

    2023

  11. [11]

    arXiv preprint arXiv:2309.17400 (2023)

    Clark, K., Vicol, P., Swersky, K., Fleet, D.J.: Directly fine-tuning diffusion models on differentiable rewards. arXiv preprint arXiv:2309.17400 (2023)

    Pith/arXiv arXiv 2023

  12. [12]

    DeepSeek-AI: Deepseek-r1: Incentivizing reasoning capability in llms via reinforce- ment learning (2025),https://arxiv.org/abs/2501.12948

    Pith/arXiv arXiv 2025

  13. [13]

    In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion

    Deng, F., Wang, Q., Wei, W., Hou, T., Grundmann, M.: Prdp: Proximal reward difference prediction for large-scale reward finetuning of diffusion models. In: Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni- tion. pp. 7423–7433 (2024)

    2024

  14. [14]

    In: 2009 IEEE conference on computer vision and pattern recognition

    Deng, J., Dong, W., Socher, R., Li, L.J., Li, K., Fei-Fei, L.: Imagenet: A large- scale hierarchical image database. In: 2009 IEEE conference on computer vision and pattern recognition. pp. 248–255. Ieee (2009)

    2009

  15. [15]

    Advances in neural information processing systems34, 8780–8794 (2021)

    Dhariwal, P., Nichol, A.: Diffusion models beat gans on image synthesis. Advances in neural information processing systems34, 8780–8794 (2021)

    2021

  16. [16]

    In: Forty-first international conference on machine learning (2024)

    Esser, P., Kulal, S., Blattmann, A., Entezari, R., Müller, J., Saini, H., Levi, Y., Lorenz, D., Sauer, A., Boesel, F., et al.: Scaling rectified flow transformers for high-resolution image synthesis. In: Forty-first international conference on machine learning (2024)

    2024

  17. [17]

    Everingham,M.,VanGool,L.,Williams,C.K.,Winn,J.,Zisserman,A.:Thepascal visualobjectclasses(voc)challenge.Internationaljournalofcomputervision88(2), 303–338 (2010)

    2010

  18. [18]

    Advances in Neural Information Processing Sys- tems36, 79858–79885 (2023) MiPO 17

    Fan, Y., Watkins, O., Du, Y., Liu, H., Ryu, M., Boutilier, C., Abbeel, P., Ghavamzadeh, M., Lee, K., Lee, K.: Dpok: Reinforcement learning for fine-tuning text-to-image diffusion models. Advances in Neural Information Processing Sys- tems36, 79858–79885 (2023) MiPO 17

    2023

  19. [19]

    In: Proceedings of the IEEE international conference on computer vision

    Fang, C., Xu, Y., Rockmore, D.N.: Unbiased metric learning: On the utilization of multiple datasets and web images for softening bias. In: Proceedings of the IEEE international conference on computer vision. pp. 1657–1664 (2013)

    2013

  20. [20]

    In: 2004 conference on computer vision and pattern recognition workshop

    Fei-Fei, L., Fergus, R., Perona, P.: Learning generative visual models from few training examples: An incremental bayesian approach tested on 101 object cate- gories. In: 2004 conference on computer vision and pattern recognition workshop. pp. 178–178. IEEE (2004)

    2004

  21. [21]

    Proceedings of the International Conference on Learning Representations (2019)

    Hendrycks,D.,Dietterich,T.:Benchmarkingneuralnetworkrobustnesstocommon corruptions and perturbations. Proceedings of the International Conference on Learning Representations (2019)

    2019

  22. [22]

    In: Proceedings of the 2021 conference on empirical methods in natural language processing

    Hessel, J., Holtzman, A., Forbes, M., Le Bras, R., Choi, Y.: Clipscore: A reference- free evaluation metric for image captioning. In: Proceedings of the 2021 conference on empirical methods in natural language processing. pp. 7514–7528 (2021)

    2021

  23. [23]

    Advances in neural information processing systems33, 6840–6851 (2020)

    Ho, J., Jain, A., Abbeel, P.: Denoising diffusion probabilistic models. Advances in neural information processing systems33, 6840–6851 (2020)

    2020

  24. [24]

    ICLR1(2), 3 (2022)

    Hu, E.J., Shen, Y., Wallis, P., Allen-Zhu, Z., Li, Y., Wang, S., Wang, L., Chen, W., et al.: Lora: Low-rank adaptation of large language models. ICLR1(2), 3 (2022)

    2022

  25. [25]

    arXiv preprint arXiv:2505.179552(2025)

    Jeong, Y., Uselis, A., Oh, S.J., Rohrbach, A.: Diffusion classifiers understand com- positionality, but conditions apply. arXiv preprint arXiv:2505.179552(2025)

    arXiv 2025

  26. [26]

    Advances in neural information processing systems36, 36652–36663 (2023)

    Kirstain,Y.,Polyak,A.,Singer,U.,Matiana,S.,Penna,J.,Levy,O.:Pick-a-pic:An open dataset of user preferences for text-to-image generation. Advances in neural information processing systems36, 36652–36663 (2023)

    2023

  27. [27]

    Springer Science & Business Media (2012)

    Kotz, S., Johnson, N.L.: Breakthroughs in Statistics: Foundations and basic theory. Springer Science & Business Media (2012)

    2012

  28. [28]

    Krizhevsky, A., Hinton, G., et al.: Learning multiple layers of features from tiny images (2009)

    2009

  29. [29]

    CS 231N7(7), 3 (2015)

    Le, Y., Yang, X.: Tiny imagenet visual recognition challenge. CS 231N7(7), 3 (2015)

    2015

  30. [30]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Li, A.C., Prabhudesai, M., Duggal, S., Brown, E., Pathak, D.: Your diffusion model is secretly a zero-shot classifier. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 2206–2217 (2023)

    2023

  31. [31]

    In: The Thirteenth International Conference on Learning Representations (2024)

    Li, A.C., Kumar, A., Pathak, D.: Generative classifiers avoid shortcut solutions. In: The Thirteenth International Conference on Learning Representations (2024)

    2024

  32. [32]

    In: European conference on computer vision

    Lin, T.Y., Maire, M., Belongie, S., Hays, J., Perona, P., Ramanan, D., Dollár, P., Zitnick, C.L.: Microsoft coco: Common objects in context. In: European conference on computer vision. pp. 740–755. Springer (2014)

    2014

  33. [33]

    arXiv preprint arXiv:2210.02747 (2022)

    Lipman, Y., Chen, R.T., Ben-Hamu, H., Nickel, M., Le, M.: Flow matching for generative modeling. arXiv preprint arXiv:2210.02747 (2022)

    Pith/arXiv arXiv 2022

  34. [34]

    arXiv preprint arXiv:2505.05470 (2025)

    Liu, J., Liu, G., Liang, J., Li, Y., Liu, J., Wang, X., Wan, P., Zhang, D., Ouyang, W.: Flow-grpo: Training flow matching models via online rl. arXiv preprint arXiv:2505.05470 (2025)

    Pith/arXiv arXiv 2025

  35. [35]

    arXiv preprint arXiv:2209.03003 (2022)

    Liu, X., Gong, C., Liu, Q.: Flow straight and fast: Learning to generate and transfer data with rectified flow. arXiv preprint arXiv:2209.03003 (2022)

    Pith/arXiv arXiv 2022

  36. [36]

    In: Proceedings of the IEEE/CVF International Conference on Computer Vision

    Ma, Y., Wu, X., Sun, K., Li, H.: Hpsv3: Towards wide-spectrum human preference score. In: Proceedings of the IEEE/CVF International Conference on Computer Vision. pp. 15086–15095 (2025)

    2025

  37. [37]

    Ain Shams engineering journal5(4), 1093–1113 (2014)

    Medhat, W., Hassan, A., Korashy, H.: Sentiment analysis algorithms and applica- tions: A survey. Ain Shams engineering journal5(4), 1093–1113 (2014)

    2014

  38. [38]

    Kim et al

    Prabhudesai, M., Goyal, A., Pathak, D., Fragkiadaki, K.: Aligning text-to-image diffusion models with reward backpropagation (2023) 18 H. Kim et al

    2023

  39. [39]

    In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., Ommer, B.: High-resolution image synthesis with latent diffusion models. In: Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. pp. 10684–10695 (2022)

    2022

  40. [40]

    International journal of computer vision 77(1), 157–173 (2008)

    Russell, B.C., Torralba, A., Murphy, K.P., Freeman, W.T.: Labelme: a database and web-based tool for image annotation. International journal of computer vision 77(1), 157–173 (2008)

    2008

  41. [41]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

    Sehwag, V., Hazirbas, C., Gordo, A., Ozgenel, F., Canton, C.: Generating high fidelity data from low-density regions using diffusion models. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. pp. 11492– 11501 (2022)

    2022

  42. [42]

    arXiv preprint arXiv:2402.03300 (2024)

    Shao, Z., Wang, P., Zhu, Q., Xu, R., Song, J., Bi, X., Zhang, H., Zhang, M., Li, Y., Wu, Y., et al.: Deepseekmath: Pushing the limits of mathematical reasoning in open language models. arXiv preprint arXiv:2402.03300 (2024)

    Pith/arXiv arXiv 2024

  43. [43]

    In: International conference on machine learning

    Sohl-Dickstein, J., Weiss, E., Maheswaranathan, N., Ganguli, S.: Deep unsuper- vised learning using nonequilibrium thermodynamics. In: International conference on machine learning. pp. 2256–2265. pmlr (2015)

    2015

  44. [44]

    arXiv preprint arXiv:2010.02502 (2020)

    Song, J., Meng, C., Ermon, S.: Denoising diffusion implicit models. arXiv preprint arXiv:2010.02502 (2020)

    Pith/arXiv arXiv 2010

  45. [45]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Stracke, N., Baumann, S.A., Bauer, K., Fundel, F., Ommer, B.: Cleandift: Diffu- sion features without noise. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 117–127 (2025)

    2025

  46. [46]

    Advances in Neural Information Processing Systems 36, 1363–1389 (2023)

    Tang, L., Jia, M., Wang, Q., Phoo, C.P., Hariharan, B.: Emergent correspon- dence from image diffusion. Advances in Neural Information Processing Systems 36, 1363–1389 (2023)

    2023

  47. [47]

    arXiv preprint arXiv:2405.18881 (2024)

    Tang, Z., Peng, J., Tang, J., Hong, M., Wang, F., Chang, T.H.: Inference- time alignment of diffusion models with direct noise optimization. arXiv preprint arXiv:2405.18881 (2024)

    Pith/arXiv arXiv 2024

  48. [48]

    In: Forty-second International Conference on Machine Learning (2025),https://openreview.net/forum?id=IH8OwjOGzM

    Um, S., Kim, B., Ye, J.C.: Boost-and-skip: A simple guidance-free diffusion for mi- nority generation. In: Forty-second International Conference on Machine Learning (2025),https://openreview.net/forum?id=IH8OwjOGzM

    2025

  49. [49]

    arXiv preprint arXiv:2301.12334 (2023)

    Um, S., Lee, S., Ye, J.C.: Don’t play favorites: Minority guidance for diffusion models. arXiv preprint arXiv:2301.12334 (2023)

    Pith/arXiv arXiv 2023

  50. [50]

    In: European Conference on Computer Vision

    Um, S., Ye, J.C.: Self-guided generation of minority samples using diffusion models. In: European Conference on Computer Vision. pp. 414–430. Springer (2024)

    2024

  51. [51]

    In: Proceedings of the Computer Vision and Pattern Recognition Conference

    Um, S., Ye, J.C.: Minority-focused text-to-image generation via prompt optimiza- tion. In: Proceedings of the Computer Vision and Pattern Recognition Conference. pp. 20926–20936 (2025)

    2025

  52. [52]

    Wallace, B., Dang, M., Rafailov, R., Zhou, L., Lou, A., Purushwalkam, S., Ermon, S., Xiong, C., Joty, S., Naik, N.: Diffusion model alignment using direct preference optimization. pp. 8228–8238 (2024)

    2024

  53. [53]

    arXiv preprint arXiv:2508.11330 (2025)

    Wang, Y., Chen, L.: Noise matters: Optimizing matching noise for diffusion clas- sifiers. arXiv preprint arXiv:2508.11330 (2025)

    arXiv 2025

  54. [54]

    In: Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers)

    Wang, Z.J., Montoya, E., Munechika, D., Yang, H., Hoover, B., Chau, D.H.: Diffu- siondb: A large-scale prompt gallery dataset for text-to-image generative models. In: Proceedings of the 61st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). pp. 893–911 (2023)

    2023

  55. [55]

    arXiv preprint arXiv:2509.20328 (2025) MiPO 19

    Wiedemer, T., Li, Y., Vicol, P., Gu, S.S., Matarese, N., Swersky, K., Kim, B., Jaini, P., Geirhos, R.: Video models are zero-shot learners and reasoners. arXiv preprint arXiv:2509.20328 (2025) MiPO 19

    Pith/arXiv arXiv 2025

  56. [56]

    arXiv preprint arXiv:2306.09341 (2023)

    Wu, X., Hao, Y., Sun, K., Chen, Y., Zhu, F., Zhao, R., Li, H.: Human preference score v2: A solid benchmark for evaluating human preferences of text-to-image synthesis. arXiv preprint arXiv:2306.09341 (2023)

    Pith/arXiv arXiv 2023

  57. [57]

    arXiv preprint arXiv:2303.144201(3) (2023)

    Wu, X., Sun, K., Zhu, F., Zhao, R., Li, H.: Better aligning text-to-image models with human preference. arXiv preprint arXiv:2303.144201(3) (2023)

    Pith/arXiv arXiv 2023

  58. [58]

    Advances in Neural Information Processing Systems36, 15903–15935 (2023)

    Xu, J., Liu, X., Wu, Y., Tong, Y., Li, Q., Ding, M., Tang, J., Dong, Y.: Imagere- ward: Learning and evaluating human preferences for text-to-image generation. Advances in Neural Information Processing Systems36, 15903–15935 (2023)

    2023

  59. [59]

    arXiv preprint arXiv:2505.07818 (2025)

    Xue, Z., Wu, J., Gao, Y., Kong, F., Zhu, L., Chen, M., Liu, Z., Liu, W., Guo, Q., Huang, W., et al.: Dancegrpo: Unleashing grpo on visual generation. arXiv preprint arXiv:2505.07818 (2025)

    Pith/arXiv arXiv 2025

  60. [60]

    Advances in Neural Information Processing Systems37, 73366–73398 (2024)

    Yuan, H., Chen, Z., Ji, K., Gu, Q.: Self-play fine-tuning of diffusion models for text-to-image generation. Advances in Neural Information Processing Systems37, 73366–73398 (2024)

    2024

  61. [61]

    In: Proceedings of the IEEE conference on computer vision and pattern recognition

    Zhang, R., Isola, P., Efros, A.A., Shechtman, E., Wang, O.: The unreasonable effectiveness of deep features as a perceptual metric. In: Proceedings of the IEEE conference on computer vision and pattern recognition. pp. 586–595 (2018)

    2018