Pith. sign in

REVIEW 3 major objections 6 minor 54 references

PRISM: Distribution-Gated Flow Matching for Controllable Unpaired Image Translation

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read PRISM replaces the single global noise knob of diffusion-based unpaired translation with a learned per-feature gate, and claims this gate, applied to both initialization and transport timing, improves realism without sacrificing structure.

desk verdict A genuinely new per-feature gate mechanism, but the baseline comparisons are too weak to support the headline FID claims. read the letter →

arxiv 2608.06240 v1 pith:W6QENRXL submitted 2026-08-06 cs.CV cs.AI

classification cs.CVcs.AI
keywords unpairedimage-to-imagetranslationflowmatchingcontrollablegenerationper-featuretransportdistributiondistancestructure-preservingmedicalimageharmonization
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper argues that the central difficulty in unpaired image-to-image translation is deciding, per image and per region, what to change and what to preserve, and that existing diffusion-based translators fail because they control this with a single global noise or guidance value. It proposes PRISM, a GAN-free method built on flow matching (a generative model that learns straight-line paths from noise to data), whose learned per-feature gate $\tau\in[0.05,1]$ is derived from each latent feature's standardized distance to the target distribution, so features already close to the target are preserved and distant features are freed. The same gate sets the ODE initial mixture $z(0)=\tau\odot z_A+(1-\tau)\odot\varepsilon_\alpha$ and the per-feature wake-up timing of transport. On five benchmarks spanning structure-preserving and structure-changing tasks, PRISM reports the best Inception FID and KID (Fréchet Inception Distance and Kernel Inception Distance) among the compared methods on four, a competitive result on the fifth, and a histopathology nuclei-count ratio closest to ideal, supporting the claimed realism-preservation balance.

What carries the argument

The load-bearing object is the per-feature preservation gate $\tau\in[0.05,1]$, one scalar per latent location, whose spatial pattern is supervised by DD$\tau$: each patch's standardized distance to the target feature distribution, quantile-normalized so distant patches receive low $\tau$ (freed) and target-like patches high $\tau$ (preserved). The gate carries the argument by being applied at both coupling points of the frozen-flow sampler: Equation (4) initializes each feature as $\tau\odot z_A+(1-\tau)\odot\varepsilon_\alpha$, mixing the real source latent with a task-matched corruption (AdaIN content anchor for structure-preserving tasks, partial isotropic anchor for structure-changing ones), and Equation (7) gates the composite velocity by the soft per-feature wake-up $\sigma((t_k-\tau)/T)$, so each feature only moves after its own threshold time. Both uses are smooth functions of $\tau$, making the gate trainable end to end against GAN-free realism objectives and a $\tau$-gated structure loss, and overridable at inference from a text prompt or a detector map without retraining.

What would settle it

Retrain PRISM on breast frozen-to-permanent translation with the gate frozen to a constant mid-value in both Equation (4) and Equation (7), keeping all losses and budget fixed; if FID and the nuclei-count ratio stay at the reported level (51.8 and 0.93), the distribution-derived spatial pattern is not what drives the balance. On the released test outputs, also check that per-feature displacement decreases monotonically with predicted $\tau$: a large fraction of high-$\tau$ features with large displacement would contradict the claimed selective transport.

Watch

Extended reading notes

Core claim

The paper's central claim is that unpaired translation can be recast as selective per-feature transport to the target distribution, controlled by a distribution-informed gate $\tau\in[\tau_{\min},\tau_{\max}]$ with the same shape as the latent. The gate's spatial prior comes from DD$\tau$, the standardized distance of each source patch to the target feature distribution under a frozen DINOv2 encoder, so target-consistent patches receive high $\tau$ and distant patches low $\tau$. The same gate enters the sampler twice: the initialization $z(0)=\tau\odot z_A+(1-\tau)\odot\varepsilon_\alpha$ mixes the real source latent with a task-matched corruption (AdaIN-style content anchoring for structure-preserving tasks, a partially anchored isotropic blend at $\alpha=0.5$ for structure-changing ones), and the integration applies a soft per-feature wake-up $\sigma((t_k-\tau)/T)$ so high-$\tau$ features stay source-anchored and low-$\tau$ features translate earlier. Over a frozen flow-matching backbone with a norm-constrained residual correction, PRISM reports the best Inception FID and KID on AFHQ cat-to-dog (76.9), CelebA-HQ man-to-woman (90.6), breast frozen-to-permanent (51.8) and day-to-night (85.9), competitive results on virtual staining (50.8 versus 49.2), and the nuclei-count ratio closest to 1.0 (0.93) on histopathology, arguing this reflects selective transport rather than under-translation.

Load-bearing premise

The load-bearing premise is that multiplying the flow's velocity by a per-feature wake-up sigmoid still keeps the trajectory on the target-image manifold, even though the gated dynamics deviate from the frozen flow's training distribution and the paper states they carry no formal distributional guarantee.

Editorial extensions

If this is right

  • On four of the five benchmarks the gated sampler achieves the best Inception FID and KID among the compared methods, with three-seed training variability ($77.4\pm0.7$ on AFHQ) well below the gap to the next-best method.
  • A single trained checkpoint can be steered locally at inference, for example by a text prompt such as 'preserve eyes' or by a nuclei-detector map, trading source preservation against target realism without any retraining.
  • Task-matched corruption avoids resynthesizing shared geometry: content-anchored initialization keeps structure while the gate transports only appearance, and on histopathology the generated-to-source nuclei-count ratio is closest to the ideal 1.0 among compared methods (0.93 versus, for example, EGSDE's 3.32).
  • The same gate field implies a per-feature compute budget: idealized active-step counts drop by 82.4% on the preservation-dominant breast task, though realizing this as wall-clock savings requires sparse execution that the paper leaves to future work.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The same checkpoint could expose the anchor strength $\alpha$ as an interactive realism-faithfulness dial at inference; the paper reports the frontier in its ablation tables but does not claim the user can change it per image after training.
  • If the wake-up schedule does what the paper argues, preserved regions should stay nearly unchanged when the number of ODE steps is increased far beyond the default of 16, because high-$\tau$ features never activate; this is a checkable consequence the paper does not state.
  • The DD$\tau$ prior is computed from a frozen encoder and target statistics only, so it could act as a plug-in preservation map for other latent-space editors beyond PRISM's own sampler, as long as those editors accept a per-feature start-point mixture.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 6 minor

Summary. PRISM is a two-stage, GAN-free flow-matching framework for unpaired image-to-image translation. Stage 1 trains a frozen domain-conditional flow; Stage 2 trains a small gate predictor, a norm-constrained correction network, and an optional learned noise generator. A per-feature gate tau, supervised by a distribution-distance prior (DDtau) computed from DINOv2 feature distances to target statistics, controls both the ODE initialization (Eq. 4) and the per-feature wake-up timing during integration (Eq. 7). The t=0 corruption is task-matched, interpolating between content-anchored AdaIN corruption and isotropic noise. Experiments on five benchmarks report the best Inception FID and KID among the compared methods on four benchmarks, a competitive result on virtual staining, and a nuclei-count ratio closest to ideal on breast histopathology, together with ablations, a realism-faithfulness analysis, and inference-time local gate overrides.

Significance. If the comparative results hold up, PRISM is a meaningful contribution: it replaces a single global noise or guidance level with a learned per-feature preservation gate that is coupled to both the initialization and the transport timing, and it does so without adversarial training. The task-matched corruption idea is well motivated, the gate-granularity ablation is well designed, and the paper is unusually candid in its limitations section, including the lack of a formal distributional guarantee for Eq. 7 and the proxy nature of the pathology metrics. The shared same-split protocol, three-seed variance studies on two benchmarks, and bootstrap uncertainty estimates are strengths. However, the central comparative claim is only as strong as the baseline implementations, and the current protocol makes the baselines weaker and less faithful than the published methods they stand for.

major comments (3)
  1. [§4, 'Baselines and protocol'; Tables 5, 7, 8, 9] The comparative conclusion 'best among evaluated methods' is under-determined by the baseline protocol. The SDEdit baseline is, in the paper's own words, 'an SDEdit-style application of our frozen Stage-1 flow rather than a reproduction of the original method', which substitutes the authors' flow for the pretrained denoiser and the per-task corruption for SDEdit's single global noise level; this is a new method, not SDEdit. EGSDE is evaluated on only two of the five benchmarks, and after retraining yields FID 109.2 on AFHQ and 171.5 on breast, values the paper acknowledges are far from its published operating point; CycleGAN and CUT also report FID values (115.0 and 162.3 on AFHQ, 74.6 on breast) outside the ranges usually associated with those methods. Because the headline claim is explicitly comparative, each baseline must be a faithful instance of the published algorithm on the same splits, or the claims must be restricted to 'our re-implementations' with the original code released. Please provide original-implementation comparisons or a sensitivity analysis showing the conclusions are robust to faithful baseline tuning, and report variance for all baselines.
  2. [§3.3 (Eq. 7) and §4.10] The second contribution, 'the same gate controls both initialization and transport timing', is not tested by any ablation that isolates Eq. 7. Table 12 varies gate granularity and Table 13 varies the DDtau prior, corruption, joint training, and correction, but no experiment keeps Eq. 4 and disables the wake-up factor in Eq. 7, or applies a global timing schedule instead of the per-feature one. Since the paper's Limitations state that the gated dynamics 'do not inherit a formal distributional guarantee from the frozen flow', an ablation of the wake-up term is the minimal experiment needed to establish that per-feature timing, rather than the gated initialization and the correction network, is responsible for the reported FID/KID improvements.
  3. [Data availability; §4 protocol] The paper states that 'the training source code, trained weights, split manifests, generated test outputs, and self-contained evaluation code can be released upon publication', but the evaluation protocol is highly custom: all baselines are retrained on split manifests that are not provided, EGSDE is retrained under a different schedule, and the FID/KID pipeline is unified. The 'best among evaluated methods' claim cannot be checked without these artifacts. Please release the code, weights, exact split definitions, and baseline configurations as part of the revision.
minor comments (6)
  1. [§3.3, Eq. (7)] At tk=1 the wake-up factor for tau=1 is sigma(0)=0.5, so a fully preserved feature still receives half the final velocity step; clarify whether this endpoint behavior is intended and whether the last step should be handled separately.
  2. [§3.5, Eq. (9)] The norm-clipping procedure says the clip is applied 'first using the channelwise norm at each spatial location and then using the norm over the full tensor'; state precisely which norm is used and in which order.
  3. [§4.2] The LNG experiment reports FID 75.0 versus 75.3 but no table or run configuration; add a table with the seed and training budget.
  4. [§4.10, Table 13] Removing the DDtau prior leaves FID essentially unchanged (103.5 vs 103.6) while changing NPS; the text should discuss explicitly why the realism metric does not move.
  5. [Abstract and §4] The qualifier 'among the evaluated methods' is used inconsistently; ensure it appears in every sentence that claims the best FID/KID, since the protocol is not directly comparable to published numbers.
  6. [Data availability] The phrase 'can be released upon publication' should be a firm commitment to release code, weights, and split manifests with links.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the per-feature gate prior is a training target, the primary evaluation metrics are external, and the self-citations are not load-bearing.

full rationale

PRISM's derivation chain is not circular in any structural sense. The DDτ prior (Eq. 12) is computed from DINOv2 feature distances to target statistics and is used as a soft regression target for the gate via L_DDτ, but this is a training objective, not a fitted parameter renamed as a prediction. The primary evaluation uses Inception FID/KID and an external H-channel nuclei-count ratio; these are not functions of the gate parameters or the DDτ prior by construction. Equations (4) and (7) implement the gate mechanism, but nothing in the objective directly optimizes the reported FID or count ratio. The paper's DINOv2-space Fréchet check is admittedly not fully independent of training because DINOv2 features appear in L_local_match and L_struct, but this is a supplementary diagnostic, not the central claim, so it does not make the main result circular. The self-citations are contextual (related work on frozen-section restoration) or forward-looking (sparse-attention future work); no uniqueness theorem is imported from the authors' prior work, and no ansatz is smuggled in by citation. The weak-baseline concern (SDEdit-style application of the authors' own frozen flow, EGSDE retrained far from published operating points) is a legitimate risk to external validity of the comparison, but it is an experimental-control issue, not a circular-reasoning issue. Overall, the central derivation is self-contained and evaluated against external benchmarks, so the appropriate finding is low circularity, with the score of 2 reflecting only the presence of minor non-load-bearing self-citations.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The method introduces no new physical or conceptual entities. The tau gate is a learned scalar field, the learned noise generator G_xi is a neural network, and the corruption types are mathematical constructions. No independent falsifiable handle outside the paper is claimed for any new entity.

free parameters (7)
  • alpha (corruption anchor strength) = 1.0 (breast, day->night), 0.5 (AFHQ, CelebA), 0.3 (virtual staining)
    Blends content-anchored AdaIN corruption with isotropic noise (Eq. 6). Values are chosen per task rather than fitted to a single dataset; Table 6 shows a strong effect on FID and LPIPS.
  • beta (correction norm bound) = 0.5 (all but virtual staining), 1.0 (virtual staining)
    Limits correction magnitude to a fraction of base velocity (Eq. 9). The per-task variation indicates hand-tuning, although the paper states training is not sensitive to beta.
  • T (wake-up sharpness) = 0.15
    Sets the sigmoid sharpness in Eq. 7, trading gradient flow against gate crispness. Fixed without a sweep.
  • c (contrast floor) = 0.85
    One-sided clamp on style standard deviation in Eq. 5 to avoid washed-out outputs on pale regions. Hand-set.
  • K (Euler steps) = 4 (training), 16 (inference)
    Integration steps; the paper notes quality is broadly stable across moderate K, so this is a mild free parameter.
  • q (DD tau quantile) = 0.95
    Normalization quantile for the distribution prior in Eq. 12. Hand-set.
  • Loss weights (lambda_mmd, lambda_DDtau, lambda_tv, lambda_spread, lambda_type, lambda_struct, etc.) = lambda_mmd=35, lambda_DDtau=0.5/1.5, lambda_tv=0.01, lambda_spread=3.0, lambda_type=3.0, lambda_struct=1.0…
    Hand-set across datasets; some weights change per regime (e.g., lambda_DDtau raised to 1.5 for Type-1 tasks). These weights shape the objective and affect results.
assumptions (5)
  • standard math Flow matching training converges to a valid generative model for each domain
    Adopts the flow-matching framework of Lipman et al. [20]; the Stage-1 model v_theta is trained with Eq. 1 and then frozen, relying on standard convergence properties.
  • domain assumption The latent space of a frozen Stable Diffusion VAE is a shared, meaningful representation for both domains
    Operates in the latent space of a frozen SD VAE (z=E(x)), C=4, H=W=32; assumes both domains map into this space and that latent distances correspond to perceptual change.
  • ad hoc to paper DINOv2 feature distance to target global moments is a valid proxy for which features should be preserved
    The DD tau prior (Eq. 12) uses each source patch's standardized distance to global target statistics (mu_B, sigma_B) to set the preservation field; this is a first-order proxy and ignores per-location conditional structure, as acknowledged in Section 3.8.
  • ad hoc to paper The gated ODE with per-feature wake-up (Eq. 7) stays near the target distribution
    The paper states in Limitations that 'the gated dynamics are an empirical design and do not inherit a formal distributional guarantee from the frozen flow.' The whole method rests on this empirical validity.
  • domain assumption Per-channel diagonal Gaussian AdaIN approximates the source-to-target transport for appearance
    Eq. 5 uses AdaIN as the content-anchored corruption, acknowledging that it 'ignores cross-channel and higher-order structure' and relies on the correction C_phi and the gate to close the gap.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PRISM: Distribution-Gated Flow Matching for Controllable Unpaired Image Translation." pith.science (2026). https://pith.science/paper/W6QENRXL

@misc{pith2026260806240,
  author       = {Pith},
  title        = {Pith review of: PRISM: Distribution-Gated Flow Matching for Controllable Unpaired Image Translation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/W6QENRXL}},
  note         = {Machine review of arXiv:2608.06240}
}
read the original abstract

Unpaired image-to-image translation must decide, per image, what to change and what to preserve without paired supervision. Many diffusion-based unpaired translators control preservation through a single global noise or guidance value applied across the image, which cannot separate content to keep from appearance to change. We present PRISM, a GAN-free flow-matching framework that replaces this global control with a learned per-feature gate. The gate's spatial prior is derived from each source feature's standardized distance to the target feature distribution, so features far from the target are freed while target-consistent features are preserved. The same gate controls both the initialization, which mixes the real source latent with a task-matched corruption, and the transport timing during Ordinary Differential Equation (ODE) integration. The corruption is matched to the task, content-anchored (AdaIN) for structure-preserving translation and partially anchored for structure-changing translation, and the gate can be overridden locally at inference from text or a detector without retraining, preserving important structures of the original image while still generating realistic results. We evaluate PRISM on five natural and biomedical benchmarks (AFHQ cat->dog, CelebA-HQ appearance translation, day->night relighting, virtual staining, and breast frozen->permanent histopathology). Among the evaluated methods under a shared same-split protocol, PRISM attains the best Inception FID and KID on four benchmarks and a competitive result on the fifth, and on histopathology yields the nuclei-count ratio closest to the ideal, supporting a favorable balance between target realism and structural preservation.

Figures

Figures reproduced from arXiv: 2608.06240 by the authors.

Figure 1
Figure 1. Overview of PRISM’s distribution-informed gated transport. The predictor Tψ produces a channel-spatial gate τ from the source latent zA, guided by a spatial prior derived from source-to-target feature-distribution discrepancy (DDτ, Section 3.8). The same gate τ enters the pipeline twice. It mixes the real source latent with the task-matched corruption ϵ (content-anchored AdaIN for structure-preserving tasks, partial… view at source ↗
Figure 2
Figure 2. Task-matched t=0 corruption (shown in pixel space for clarity, computed on the latent). Columns, left to right. The source zA. Instance normalization IN(zA), which retains spatial structure and drops global colour. A target sample supplying the per-channel style (µ, ˜ σ˜). The content-anchored corruption ϵstruct, source structure with target style, whose τ→0 fixed point is a valid style transfer, and the isotropic n… view at source ↗
Figure 4
Figure 4. Qualitative comparison on AFHQ cat→dog (held-out test cats, one per row). Columns, left to right: input cat, PRISM (ours), CycleGAN, CUT, UNSB, SDEdit. PRISM produces realistic dogs that inherit the source cat’s colour, pose and gaze, whereas CUT under-translates (the output stays cat-like), CycleGAN retains cat features, and SDEdit introduces artifacts [PITH_FULL_IMAGE:figures/full_fig_p011_4.png] view at source ↗
Figures from the paper (8 more)
Figure 5
Figure 5. Figure 5: Qualitative comparison on CelebA-HQ man→woman (held-out test, one source per row). Columns, left to right: input, PRISM, CycleGAN, CUT, UNSB, SDEdit. PRISM alters the target-labelled facial attributes while retaining visually similar pose and lighting. CUT changes the …
Figure 6
Figure 6. Figure 6: Qualitative breast frozen→permanent on reassembled full sections (held-out test, one section per row, tiles stitched with a shared corruption seed and blended over a 25% overlap so no seams appear). Columns, left to right: input frozen section, PRISM, CycleGAN, CUT, UN…
Figure 9
Figure 9. Figure 9: presents this analysis for one benchmark from each regime. On breast frozen→permanent transla￾tion, PRISM combines the lowest Inception FID with the nuclei-count ratio closest to the ideal value of 1.0. Other methods either remain insufficiently translated or intro￾duc…
Figure 7
Figure 7. Figure 7: Qualitative comparison on day→night. Columns, left to right: input day image, PRISM, CycleGAN, CUT, UNSB, SDEdit. PRISM relights to a convincing night while holding scene geometry largely fixed. UNSB and SDEdit shift structure or wash out detail [PITH_FULL_IMAGE:figur…
Figure 8
Figure 8. Figure 8: Qualitative comparison on virtual staining (unstained [PITH_FULL_IMAGE:figures/full_fig_p015_8.png]
Figure 10
Figure 10. Figure 10: Text-guided local preservation during AFHQ cat [PITH_FULL_IMAGE:figures/full_fig_p016_10.png]
Figure 9
Figure 9. Figure 9: Joint realism-faithfulness analysis. The vertical axis [PITH_FULL_IMAGE:figures/full_fig_p016_9.png]
Figure 11
Figure 11. Figure 11: Relationship between predicted gate values and latent [PITH_FULL_IMAGE:figures/full_fig_p017_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

54 extracted references · 51 canonical work pages

  1. [1]

    Isola, J.-Y

    P. Isola, J.-Y. Zhu, T. Zhou, A. A. Efros, Image-to- image translation with conditional adversarial net- works, in: IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition, 2017

  2. [2]

    J.-Y. Zhu, T. Park, P. Isola, A. A. Efros, Unpaired image-to-image translation using cycle-consistent ad- versarial networks, in: IEEE/CVF International Con- ference on Computer Vision, 2017

  3. [3]

    T. Park, A. A. Efros, R. Zhang, J.-Y. Zhu, Con- trastive learning for unpaired image-to-image trans- lation, in: European Conference on Computer Vision, 2020

  4. [4]

    Goodfellow, J

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, Y. Bengio, Generative adversarial nets, in: Advances in Neural Information Processing Systems, 2014

  5. [5]

    C. Meng, Y. He, Y. Song, J. Song, J. Wu, J.-Y. Zhu, S. Ermon, SDEdit: Guided image synthesis and editing with stochastic differential equations, in: In- ternational Conference on Learning Representations, 2022

  6. [6]

    Huang, M.-Y

    X. Huang, M.-Y. Liu, S. Belongie, J. Kautz, Multi- modal unsupervised image-to-image translation, in: European Conference on Computer Vision, 2018

  7. [7]

    Y. Choi, Y. Uh, J. Yoo, J.-W. Ha, StarGAN v2: Diverse image synthesis for multiple domains, in: IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, 2020

  8. [8]

    H. Tang, H. Liu, D. Xu, P. H. S. Torr, N. Sebe, AttentionGAN: Unpaired image-to-image transla- tion using attention-guided generative adversarial networks, IEEE Transactions on Neural Networks and Learning Systems 34 (4) (2023) 1972–1987. doi:10.1109/TNNLS.2021.3105725

Show all 54 references
  1. [9]

    Stuhr, J

    B. Stuhr, J. Brauer, B. Schick, J. Gonzàlez, Masked discriminators for content-consistent unpaired image- to-image translation, IEEE Access 14 (2026) 2486– 2509.doi:10.1109/ACCESS.2024.3488837

  2. [10]

    J. Ho, A. Jain, P. Abbeel, Denoising diffusion proba- bilistic models, in: Advances in Neural Information Processing Systems, 2020

  3. [11]

    Y. Song, J. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, B. Poole, Score-based generative modeling through stochastic differential equations, in: Inter- national Conference on Learning Representations, 2021

  4. [12]

    J. Song, C. Meng, S. Ermon, Denoising diffusion im- plicit models, in: International Conference on Learn- ing Representations, 2021

  5. [13]

    Rombach, A

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, B. Ommer, High-resolution image synthesis with latent diffusion models, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022. 18

  6. [14]

    Zhang, A

    L. Zhang, A. Rao, M. Agrawala, Adding condi- tional control to text-to-image diffusion models, in: IEEE/CVF International Conference on Computer Vision, 2023

  7. [15]

    M. Zhao, F. Bao, C. Li, J. Zhu, EGSDE: Unpaired image-to-imagetranslationviaenergy-guidedstochas- tic differential equations, in: Advances in Neural Information Processing Systems, 2022

  8. [16]

    S. Sun, L. Wei, J. Xing, J. Jia, Q. Tian, SDDM: Score-decomposed diffusion models on manifolds for unpaired image-to-image translation, in: Interna- tional Conference on Machine Learning, 2023

  9. [17]

    B. Kim, J. C. Ye, Unpaired image-to-image transla- tion via neural Schrödinger bridge, in: International Conference on Learning Representations, 2024

  10. [18]

    Y. Shi, V. De Bortoli, A. Campbell, G. Deligiannidis, A. Arnsten, Diffusion Schrödinger bridge matching, in: Advances in Neural Information Processing Sys- tems, 2024

  11. [19]

    Parmar, T

    G. Parmar, T. Park, S. Narasimhan, J.-Y. Zhu, One- step image translation with text-to-image models, arXiv preprint arXiv:2403.12036 (2024)

  12. [20]

    Lipman, R

    Y. Lipman, R. T. Chen, H. Ben-Hamu, M. Nickel, Flow matching for generative modeling, in: Inter- national Conference on Learning Representations, 2023

  13. [21]

    X. Liu, C. Gong, Q. Liu, Flow straight and fast: Learning to generate and transfer data with recti- fied flow, in: International Conference on Learning Representations, 2023

  14. [22]

    A. Tong, N. Malkin, G. Huguet, Y. Zhang, J. Rector- Brooks, K. Fatras, G. Wolf, Y. Bengio, Improving and generalizing flow-based generative models with minibatch optimal transport, Transactions on Ma- chine Learning Research (2024)

  15. [23]

    J. Choi, S. Kim, Y. Jeong, Y. Gwon, S. Yoon, ILVR: Conditioning method for denoising diffusion proba- bilistic models, in: IEEE/CVF International Confer- ence on Computer Vision, 2021

  16. [24]

    Avrahami, D

    O. Avrahami, D. Lischinski, O. Fried, Blended dif- fusion for text-driven editing of natural images, in: IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, 2022

  17. [25]

    G. Kwon, J. C. Ye, Diffusion-based image translation using disentangled style and content representation, in: International Conference on Learning Represen- tations, 2023

  18. [26]

    X. Su, J. Song, C. Meng, S. Ermon, Dual diffusion implicit bridges for image-to-image translation, in: International Conference on Learning Representa- tions, 2023

  19. [27]

    G.-H. Liu, A. Vahdat, D.-A. Huang, E. A. Theodorou, W. Nie, A. Anandkumar, I2SB: Image-to-image Schrödinger bridge, in: International Conference on Machine Learning, 2023

  20. [28]

    Couairon, J

    G. Couairon, J. Verbeek, H. Schwenk, M. Cord, DiffEdit: Diffusion-based semantic image editing with mask guidance, in: International Conference on Learning Representations, 2023

  21. [29]

    T. C. Hollon, B. Pandian, A. R. Adapa, E. Urias, A. V. Save, S. S. S. Khalsa, D. G. Eichberg, R. S. D’Amico, Z. U. Farooq, S. Lewis, et al., Deep learning for intraoperative frozen section diagnosis of brain tumors, Nature Medicine 26 (2020) 52–58

  22. [30]

    de Haan, Y

    K. de Haan, Y. Zhang, J. E. Zuckerman, T. Liu, A. Deng, Y. Jia, J. Li, Y. Rivenson, A. Ozcan, Deep learning-based transformation of H&E stained tissues into special stains, Nature Communications 12 (2021) 4884

  23. [31]

    Rivenson, H

    Y. Rivenson, H. Wang, Z. Wei, K. de Haan, Y. Zhang, Y. Wu, H. Günaydın, J. E. Zuckerman, T. Chong, A. E. Sisk, et al., Virtual histological staining of unlabelled tissue autofluorescence images via deep learning, Nature Biomedical Engineering 3 (2019) 466–477.doi:10.1038/s4155...

  24. [32]

    J. Y. You, M. Eom, T.-I. Choi, E.-S. Cho, J. Choi, M. Lee, C. Shin, J. Moon, E. Kim, P. Kim, C.-H. Kim, Y.-G. Yoon, Preserving spatial and quantita- tive information in unpaired biomedical image-to- image translation, Cell Reports Methods 5 (6) (2025) 101074.doi:10.1016/j.crme...

  25. [33]

    K. B. Özyörük, S. Can, B. Darbaz, K. Basak, D. Demir, G. I. Gokceler, K. Sali, C. Gunduz-Demir, A deep-learning model for transforming the style of tissue images from cryosectioned to formalin- fixed and paraffin-embedded, Nature Biomedical Engineering 6 (2022) 1407–1419. doi:...

  26. [34]

    J. J. Levy, C. C. Haudenschild, C. Barber, L. Chark- oudian, L. J. Vaickus, A large-scale study of histopathological stain transfer with generative mod- els, Journal of Pathology Informatics (2024)

  27. [35]

    Yoshai, G

    E. Yoshai, G. Goldinger, T. Kogan, A. Zakharov, M. Haifler, N. T. Shaked, Enhancing frozen his- tological section images using permanent-section- guided deep learning with nuclei attention, Scien- tific Reports 15 (1) (2025) 30594. doi:10.1038/ s41598-025-12181-6. 19

  28. [36]

    Filiot, R

    A. Filiot, R. Gherber, A. Dehaene, C. Dez, A. Bussy, T. Courtin, E. Picard, I. Usicenko, P. Courtiol, Scal- ing self-supervised learning for histopathology with masked image modeling, in: Medical Image Comput- ing and Computer-Assisted Intervention, 2023

  29. [37]

    Filiot, R

    A. Filiot, R. Gherber, A. Dehaene, C. Dez, A. Bussy, T. Courtin, E. Picard, I. Usicenko, P. Cour- tiol, Phikon-v2: A large-scale vision foundation model for computational pathology, arXiv preprint arXiv:2409.09818 (2024)

  30. [38]

    R. J. Chen, T. Ding, M. Y. Lu, D. F. Williamson, G. Jaume, A. H. Song, B. Chen, A. Zhang, D. Shao, M. Shaban, et al., Towards a general-purpose foun- dation model for computational pathology, Nature Medicine 30 (2024) 850–862

  31. [39]

    M. Y. Lu, B. Chen, D. F. Williamson, R. J. Chen, I. Liang, T. Ding, G. Jaume, I. Williamson, A. Zhang, F. Mahmood, A visual-language foundation model for computational pathology, Nature Medicine 30 (2024) 863–874

  32. [40]

    The Cancer Genome Atlas Research Network, The cancer genome atlas pan-cancer analysis project, Na- ture Genetics 45 (2013) 1113–1120

  33. [41]

    Huang, S

    X. Huang, S. Belongie, Arbitrary style transfer in real-time with adaptive instance normalization, in: IEEE/CVF International Conference on Computer Vision, 2017

  34. [42]

    Peebles, S

    W. Peebles, S. Xie, Scalable diffusion models with transformers, in: IEEE/CVF International Confer- ence on Computer Vision, 2023

  35. [43]

    Oquab, T

    M. Oquab, T. Darcet, T. Moutakanni, H. Vo, M. Szafraniec, V. Khalidov, P. Fernandez, D. Haziza, F. Massa, A. El-Nouby, et al., DINOv2: Learning robust visual features without supervision, in: Trans- actions on Machine Learning Research, 2024

  36. [44]

    2021 (2021)

    Steve Mark, Day-night dataset, Kaggle, https://www.kaggle.com/datasets/stevemark/ daynight-dataset, last updated Apr. 2021 (2021)

  37. [45]

    Z. Asaf, A. Abdul Salam, B. Rao, S. Khan, U. Akram, B. Sanabria, E-staining DermaRepo: Paired unstained and H&E-stained dermatopathol- ogy whole slide images, https://data.mendeley. com/datasets/gxgg933ny3/1, version 1 (2024). doi:10.17632/gxgg933ny3.1

  38. [46]

    Karras, T

    T. Karras, T. Aila, S. Laine, J. Lehtinen, Progressive growing of GANs for improved quality, stability, and variation, in: International Conference on Learning Representations, 2018

  39. [47]

    Heusel, H

    M. Heusel, H. Ramsauer, T. Unterthiner, B. Nessler, S. Hochreiter, GANs trained by a two time-scale update rule converge to a local Nash equilibrium, in: Advances in Neural Information Processing Systems, 2017

  40. [48]

    Bińkowski, D

    M. Bińkowski, D. J. Sutherland, M. Arbel, A. Gret- ton, Demystifying MMD GANs, in: International Conference on Learning Representations, 2018

  41. [49]

    Schmidt, M

    U. Schmidt, M. Weigert, C. Broaddus, G. Myers, Cell detection with star-convex polygons, in: Medical Im- age Computing and Computer-Assisted Intervention, 2018

  42. [50]

    Zhang, P

    R. Zhang, P. Isola, A. A. Efros, E. Shechtman, O. Wang, The unreasonable effectiveness of deep features as a perceptual metric, in: IEEE/CVF Con- ference on Computer Vision and Pattern Recognition, 2018

  43. [51]

    Parmar, R

    G. Parmar, R. Zhang, J.-Y. Zhu, On aliased resiz- ing and surprising subtleties in gan evaluation, in: IEEE/CVF Conference on Computer Vision and Pat- tern Recognition, 2022

  44. [52]

    J. Ho, T. Salimans, Classifier-free diffusion guidance, arXiv preprint arXiv:2207.12598 (2022)

  45. [53]

    Lüddecke, A

    T. Lüddecke, A. Ecker, Image segmentation using text and image prompts, in: IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2022

  46. [54]

    Yoshai, D

    E. Yoshai, D. Yagoda-Aharoni, E. Dotan, N. T. Shaked, Hierarchical sparse attention framework for computationally efficient classification of biological cells, arXiv preprint arXiv:2505.07661 (2025).arXiv: 2505.07661,doi:10.48550/arXiv.2505.07661. 20

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.