Pith. sign in

REVIEW 5 major objections 5 minor 56 references

DiffEx: Explaining a Classifier with Diffusion Models to Identify Microscopic Cellular Variations

T0 review · 5 major / 5 minor · reviewed 2026-08-08 · deepseek-v4-flash

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

Pith's one-line read DiffEx generates visually interpretable attributes that explain a classifier and reveal microscopic cellular variations between conditions.

desk verdict DiffEx is a promising classifier-explanation pipeline, but the current version doesn't rule out that its discovered directions are partly artifacts of moving the classifier score through the concatenated latent code. read the letter →

arxiv 2502.09663 v1 pith:HQQETFZF submitted 2025-02-12 cs.CV cs.AIcs.LGq-bio.CB

Anis Bourou , Saranga Kingkor Mahanta , Thomas Boyer , Valérie Mezger , Auguste Genovesio This is my paper · ORCID
classification cs.CVcs.AIcs.LGq-bio.CB
keywords diffusionmodelsclassifierexplainabilitylatentspacedirectionscontrastivelearningcellularphenotypediscoverymicroscopyimageanalysiscounterfactualexplanationssemantic
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

DiffEx is a method for explaining what a trained image classifier sees, by generating a ranked set of visual attributes that push the classifier from one class to another. The paper claims that these attributes are not just heatmaps but editable image directions: shifting an image's latent code along a learned direction changes one interpretable property at a time, such as lipstick, hairstyle, cytoplasm loss, or Golgi scattering. On two microscopy datasets, the top-ranked directions replicate known treatment phenotypes, including cell death and cytoskeletal disruption on BBBC021 and Nocodazole-induced Golgi scattering on the Golgi dataset. If correct, the method turns a black-box classifier into a phenotyping tool that could suggest new biomarkers and treatment effects without hand-crafted image features.

What carries the argument

The load-bearing object is the classifier-aware semantic code $z_{\mathrm{sem}}$, formed by concatenating an encoder's latent vector with the classifier's prediction score and used to condition the DDIM denoiser, which makes the latent space carry classifier-relevant attributes. Direction discovery is carried by MLP-based direction models $D_k(z,\alpha)=z+\alpha\,\mathrm{MLP}_1(z)/\|\mathrm{MLP}_1(z)\|$, whose edits are separated by a contrastive loss (Eq. 8) and decorrelated by a covariance regularization term (Eq. 9), then ranked by average change in classification score. The mechanism's job is to turn the classifier's decision surface into a short list of global, one-at-a-time attribute edits.

What would settle it

Apply the top-ranked direction from the treated-versus-untreated classifier to a held-out set of untreated images and measure the target phenotype (e.g., cytoplasm area, nuclei count, Golgi area) automatically: the claim fails if the direction does not change the measured phenotype in the expected direction for most images, or if it also changes unrelated attributes.

Watch

Extended reading notes

Core claim

The central claim is that a classifier can be explained by training a diffusion autoencoder whose semantic latent code includes the classifier's score, then learning a set of direction models $D_k(z,\alpha)=z+\alpha\,\mathrm{MLP}_1(z)/\|\mathrm{MLP}_1(z)\|$ that shift the latent code along normalized directions, with a contrastive loss that makes edits from the same direction attract and edits from different directions repel. The discovered directions are ranked by how much they change the classifier's probability, and applying a direction to an image produces a counterfactual that visibly exhibits a single attribute. The paper shows directions for gender classification (lipstick, hairstyle length) and, on biological data, directions that replicate the phenotypes induced by the tested compounds: cytoplasm disappearance, nuclei reduction, red-channel intensity changes, and nuclei clustering for Latrunculin B, and Golgi apparatus scattering for Nocodazole. The ranked directions are meant to be the classifier's own decision attributes, made visible and editable.

Load-bearing premise

The paper assumes that a single shift of the latent code moves the same visual attribute in every image, so a direction learned from some images will edit the same attribute in others.

Editorial extensions

If this is right

  • Applied to treated versus untreated microscopy images, the top-ranked directions recover multiple known drug-induced phenotypes, including cytoplasm loss, nuclei reduction, and red-channel intensity changes on BBBC021.
  • On the Golgi dataset, the identified direction reproduces Nocodazole-induced Golgi scattering, and the reverse direction reproduces aggregation.
  • On the FFHQ gender classifier, directions correspond to recognizable attributes such as lipstick and hairstyle length, showing the method works on natural images as well.
  • Because reconstruction keeps classifier accuracy high, the edited images are not adversarial artifacts but preserve the features the classifier relies on.
  • The ranking procedure yields a short list of directions that most affect the classifier, rather than requiring the user to inspect all latent dimensions.

Reading between the lines

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

  • Because ranking removes images already explained, the number of useful directions is bounded by the diversity of the image sample, which may explain why the Golgi experiment collapsed onto a single phenotype.
  • Deployed prospectively on an uncharacterized drug screen, the top-ranked directions would generate candidate biomarkers that could be tested with targeted stains or single-cell assays.
  • Making the direction models conditional on the starting class, or on multiple classifier scores, might separate additional attributes in datasets where directions currently collapse.
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

5 major / 5 minor

Summary. DiffEx proposes to explain image classifiers by learning a set of interpretable directions in a semantic latent space built with a diffusion autoencoder. The semantic code concatenates an encoder embedding with the classifier score and is used to condition a DDIM, while a KL classifier loss encourages reconstructions to preserve the score. Direction models are then trained with a contrastive loss and a decorrelation regularizer, and directions are ranked by the average classifier-score shift they induce. The authors demonstrate the method on FFHQ gender classification and on two fluorescence microscopy datasets (BBBC021 and Golgi), claiming that the top directions replicate known treatment phenotypes such as cytoplasm loss, nuclear count reduction, and Golgi scattering.

Significance. If the central claims hold, DiffEx would be a useful interpretability tool: it converts a black-box classifier into a ranked set of visually editable attribute directions and can potentially surface subtle phenotypes in microscopy data. The idea of making the latent space classifier-aware and then discovering directions with contrastive learning is a reasonable extension of DiffAE and NoiseCLR. However, the current evidence is mostly qualitative: phenotype replication is asserted from selected figures, disentanglement is not measured, and the classifier-score concatenation introduces a confounding channel that is never analyzed. With additional ablations and quantitative evaluation on generated images, the contribution could be significant for both interpretability and bioimage analysis; as written, the load-bearing claims are not yet established.

major comments (5)
  1. [§3.1, Eq. (2), Eq. (4), §3.3] The concatenation of the classifier score into zsem creates a confound that is not addressed. The direction model Dk(z, α) in Eq. (4) is free to move the score coordinate of zsem, and the ranking procedure in §3.3 explicitly retains directions whose average classifier-score change is large. Since the DDIM is conditioned on zsem, changing the score coordinate alone can produce target-class-like appearances; the selected directions may therefore be score-conditional sampling artifacts rather than attribute directions in the encoder's semantic latent space. The authors should report, for each discovered direction, the proportion of its norm that lies in the score coordinate(s), or ablate by projecting MLP1(z) orthogonal to the score subspace, or compare against a baseline that changes only the score coordinate. Without one of these analyses, the explanation claim is not established.
  2. [§4.3, Figs. 4–6] The phenotype-replication claims are supported only by visual inspection of generated images. Fig. 6 reports CellProfiler measurements on real images only, so it confirms that the two conditions differ in the expected way but does not show that DiffEx's shifted images exhibit those differences. The authors should measure the same morphological features on the generated/shifted images and report, for each direction, whether the feature moves in the claimed direction and by how much relative to the real-condition gap. Without such measurements, statements such as 'Direction 1 eliminates the cytoplasm' or 'the Golgi apparatus becomes more scattered' remain qualitative assertions.
  3. [§3.2, Eq. (8), §4.3] The claim that the discovered directions are 'well disentangled and separated' is not backed by a quantitative metric. Equation (8) is a ranking loss over feature differences computed by MLP2; it may separate the feature representations used in the contrastive objective, but it does not by itself ensure that shifting along one discovered direction leaves the attributes associated with other directions invariant. The authors should provide a quantitative disentanglement evaluation, for example by measuring attribute classifier outputs or pixel-level feature changes during traversal along each direction and computing pairwise interference.
  4. [§3.2–§3.3; Supplementary B] Several central hyperparameters and implementation details are missing: λ1, λ2, α, the number of directions N, the ranking threshold, the number of images in the ranking sample, and the architecture/training details of the two MLPs. The ranking pseudo-code is deferred to Supplementary B, but the supplementary material is not included with the manuscript. Since the ranking threshold determines which directions are retained and α controls the edit magnitude, the experiments cannot be reproduced or fully assessed as presented.
  5. [§3.2, Eq. (8)] The denominator of Eq. (8) contains only negative-pair terms, making the loss -log(pos/neg) rather than a standard InfoNCE loss whose denominator also includes positive pairs. Please clarify whether this is intentional and define the exact sets of positive and negative pairs in both words and equations, including the role of the self-pair (j = i) and the treatment of terms with l = k.
minor comments (5)
  1. [End of §3.2] In Eq. (10), the text says 'where λ1 is a hyperparameter', but the expression contains λ2; please correct this to λ2 (or state the intended weighting).
  2. [§4.4] The comparison section refers to 'In Fig. 4, we further evaluate...' but the GCD comparison appears in Fig. 8; also, the sentence about CellProfiler contains 'As shown in Fig.' with a missing figure number.
  3. [Eq. (4)] The direction-specific network is written as MLP1 without a subscript k; please use MLP1^k (or similar) to make explicit that each direction model has its own parameters.
  4. [Table 2] KID is computed on fluorescence microscopy images using an ImageNet-pretrained Inception network, which may not be well suited to this domain; please report confidence intervals and consider domain-specific embeddings or FID variants.
  5. [References] Several citations are malformed, including 'et al, 2006', 'et al., 2022', and 'et al., 2010', which should be given proper author names; in addition, the supplementary material (Sections A and B) is referenced but not included.

Circularity Check

1 steps flagged · score 6.0 of 10

DiffEx's discovered directions are ranked by classifier-score change while zsem directly contains the classifier score, so the explanations may reduce to score-conditioned generation rather than latent attribute discovery.

  1. self definitional [Section 3.1 (Eq. 1), Section 3.2 (Eq. 4), Section 3.3]
    "To ensure that the generated samples maintain classifier-relevant attributes, we concatenate the classification score with the latent vector, forming a semantic code to condition the diffusion model, we denote it as zsem. ... Dk(z, α) = z + α MLP1(z)/∥MLP1(z)∥ (4) ... For each discovered direction, we shift all images in the sample along that direction by a specific value of α and then calculate the new classification scores for the shifted images. If the average change in classification scores exceeds a predefined threshold, we retain that direction."

    The semantic code zsem includes the classifier score as a conditioning component, and the direction model D_k in Eq. 4 is free to shift any component of zsem, including the score channel, because no constraint is imposed on MLP1's output support. The ranking step then retains exactly those directions whose shift changes the classifier score. Hence any direction that moves the score component is guaranteed to pass selection, and the DDIM generates the resulting image conditioned on the altered score. The 'interpretable attribute' is therefore, by construction, confounded with a direct score manipulation: DiffEx can produce the target-class appearance encoded in the generator rather than an attribute direction discovered in the encoder's semantic latent space.

full rationale

The central concern is that DiffEx builds its semantic code by concatenating the classifier's score with the encoder latent (Eq. 1), then learns and ranks directions by how much they change that same classifier score (Section 3.3). Because Eq. 4 does not restrict the direction model from altering the score component, the selected explanations may be generated by simply conditioning the diffusion model on an altered score, which is a class-conditional artifact rather than a discovered visual attribute. This is the one substantive circularity: the explanation pipeline is self-referential by design, and the paper does not disentangle score-subspace shifts from genuinely semantic latent shifts. The biological phenotype validations in Section 4.3 are checked against known external phenotypes (e.g., Golgi scattering) and were not used to fit the direction models, so those findings retain independent content and are not themselves circular. The self-citations to the authors' prior CycleGAN and PhenDiff work appear only as background and are not load-bearing. No uniqueness theorem or ansatz is smuggled in via citation. Overall, the paper has one central design-level circularity that partially undermines the claim that the discovered directions explain the classifier through latent attributes, but the external phenotype evaluation and the contrastive-learning component provide independent content, warranting a score of 6 rather than higher.

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

The central claim rests on the assumption that a linear shift in a learned latent space yields disentangled attributes, plus the standard machinery of diffusion autoencoders and contrastive learning. No new physical entities are introduced. The free parameters are the hyperparameters that the paper does not report.

free parameters (5)
  • lambda_1 = not reported
    Weight of the classifier KL loss L_cls in L_sem = L_diffusion + lambda_1 L_cls (Eq. 3). It controls how strongly the latent space preserves classifier scores.
  • lambda_2 = not reported
    Weight of the decorrelation regularizer in L_dir = L_cont + lambda_2 L_reg (Eq. 10, mislabeled lambda_1 in the text). It controls direction disentanglement.
  • alpha = not reported
    Shift magnitude applied along discovered directions in D_k(z, alpha) = z + alpha MLP1(z)/||MLP1(z)|| (Eq. 4). Affects visual strength and classifier shift.
  • N (number of directions) = 10 in supplementary, 3 shown in figures
    Number of directional models learned. Chosen by the user; not justified.
  • ranking threshold = not reported
    Predefined threshold on average classifier score change used to retain directions (Section 3.3). Not specified in the text.
assumptions (5)
  • standard math Diffusion models can reconstruct images from a semantic latent code via DDIM sampling, and the encoder learns a meaningful decodable representation (DiffAE assumption).
    Used to build the semantic space in Section 3.1, following Preechakul et al. 2022.
  • domain assumption A linear shift in the learned semantic latent space, z + alpha MLP1(z)/||MLP1(z)||, produces a single interpretable attribute change that generalizes across images.
    Core assumption of direction models in Section 3.2, Eq. 4. If false, the discovered directions are not meaningful explanations.
  • domain assumption The classifier's class probabilities are a sufficient conditioning signal to preserve classifier-relevant attributes in the reconstructed image.
    Used in L_cls (Eq. 2) and for zsem conditioning. Assumes probabilities capture the attributes driving classification.
  • domain assumption The contrastive loss over feature differences f_k separates directions so that edits from the same direction are similar and edits from different directions are distinct.
    Standard contrastive learning assumption applied in Section 3.2, Eq. 8.
  • domain assumption The trained classifiers are accurate proxies for the biological conditions, so explaining them yields biologically meaningful phenotypes.
    Accuracy is high in Table 1, but the mapping from classifier features to true biology is assumed, not proven.

how reviews work

0 comments
Cite this review

Pith. "Pith review of DiffEx: Explaining a Classifier with Diffusion Models to Identify Microscopic Cellular Variations." pith.science (2026). https://pith.science/paper/HQQETFZF

@misc{pith2026250209663,
  author       = {Pith},
  title        = {Pith review of: DiffEx: Explaining a Classifier with Diffusion Models to Identify Microscopic Cellular Variations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HQQETFZF}},
  note         = {Machine review of arXiv:2502.09663}
}
read the original abstract

In recent years, deep learning models have been extensively applied to biological data across various modalities. Discriminative deep learning models have excelled at classifying images into categories (e.g., healthy versus diseased, treated versus untreated). However, these models are often perceived as black boxes due to their complexity and lack of interpretability, limiting their application in real-world biological contexts. In biological research, explainability is essential: understanding classifier decisions and identifying subtle differences between conditions are critical for elucidating the effects of treatments, disease progression, and biological processes. To address this challenge, we propose DiffEx, a method for generating visually interpretable attributes to explain classifiers and identify microscopic cellular variations between different conditions. We demonstrate the effectiveness of DiffEx in explaining classifiers trained on natural and biological images. Furthermore, we use DiffEx to uncover phenotypic differences within microscopy datasets. By offering insights into cellular variations through classifier explanations, DiffEx has the potential to advance the understanding of diseases and aid drug discovery by identifying novel biomarkers.

Figures

Figures reproduced from arXiv: 2502.09663 by the authors.

Figure 1
Figure 1. DiffEx primarily consists of three stages: (a) A semantic latent space is constructed by combining the embedding obtained from an encoder with the classifier’s prediction for each image. The resulting representation is used to condition the DDIM. (b) Directional models are learned in this semantic latent space using a self-supervised approach. (c) After identifying the directions that most significantly affect the c… view at source ↗
Figure 2
Figure 2. Shifting images toward the opposite class using directions identified by Diffex. Left: When transforming male images toward the female class, the appearance of lipstick becomes noticeable, suggesting it as a discriminative attribute for the classifier. Right: When shifting female images toward the male class, hairstyles tend to become shorter, indicating an attribute associated with the male class. The probabilities… view at source ↗
Figure 3
Figure 3. Images from two datasets: (a) BBBC021 dataset and (b) Golgi dataset. While the differences between the two classes are apparent in BBBC021—such as the disappearance of the cyto￾plasm and fewer nuclei—they are more subtle in the Golgi dataset. to the numerator of the loss function. Conversely, feature divergences originating from different directional models (e.g., f k 1 ̸= f l 1 , , l ̸= k) are treated as negative p… view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Shifting images toward the opposite class. Left: DiffEx identified three distinct directions for transitioning from the untreated to the treated class. Direction 1 eliminates the cytoplasm and most cells, leaving a single nucleus at the center. Direction 2 removes the …
Figure 5
Figure 5. Figure 5: Shifting images toward the opposite class. Left: When transitioning from the treated to the untreated class, the Golgi ap￾paratus tends to aggregate. Right: Conversely, shifting from the untreated to the treated class results in its dispersion. These ob￾servations repl…
Figure 6
Figure 6. Figure 6: Left: Measurement of the Golgi apparatus area in real images for both conditions reveals a difference in its spatial dis￾tribution. The area is larger in the treated case due to treatment￾induced scattering. Right: Measurement of the nuclear area in the BBBC021 dataset…
Figure 7
Figure 7. Figure 7: Generating counterfactual explanation with our method and GCD. We can see that our method gives visually better and more disentangled results. and more disentangled compared to those obtained using GCD. Specifically, our method focus on modifying a single attribute—pri…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 32 canonical work pages

  1. [1]

    Diffusion visual counterfactual explanations

    Augustin, M., Boreiko, V., Croce, F., and Hein, M. Diffusion visual counterfactual explanations. In Advances in Neural Information Processing Systems (NeurIPS), 2022

    work page 2022

  2. [2]

    VICReg : Variance-invariance-covariance regularization for self-supervised learning

    Bardes, A., Ponce, J., and LeCun, Y. VICReg : Variance-invariance-covariance regularization for self-supervised learning. In International Conference on Learning Representations, 2022. URL https://openreview.net/forum?id=xm6YD62D1Ub

    work page 2022

  3. [3]

    J., Arbel, M., and Gretton, A

    Bińkowski, M., Sutherland, D. J., Arbel, M., and Gretton, A. Demystifying mmd gans, 2021. URL https://arxiv.org/abs/1801.01401

    arXiv 2021

  4. [4]

    and Genovesio, A

    Bourou, A. and Genovesio, A. Unpaired image-to-image translation with limited data to reveal subtle phenotypes, 2023

    work page 2023

  5. [5]

    PhenDiff: Revealing Subtle Phenotypes with Diffusion Models in Real Images

    Bourou, A., Boyer, T., Gheisari, M., Daupin, K., Dubreuil, V., De Thonel, A., Mezger, V., and Genovesio, A. PhenDiff: Revealing Subtle Phenotypes with Diffusion Models in Real Images . In proceedings of Medical Image Computing and Computer Assisted Intervention -- MICCAI 2024, volume LNCS 15003. Springer Nature Switzerland, October 2024

    work page 2024

  6. [6]

    Large scale GAN training for high fidelity natural image synthesis

    Brock, A., Donahue, J., and Simonyan, K. Large scale GAN training for high fidelity natural image synthesis. In International Conference on Learning Representations (ICLR), 2019. URL https://openreview.net/forum?id=B1xsqj09Fm

    work page 2019

  7. [7]

    Chandrasekaran, S. N. C., Ceulemans, H., Boyd, J. D., and Carpenter, A. E. Image-based profiling for drug discovery: due for a machine-learning upgrade? Nature Reviews Drug Discovery, 20: 0 145--159, 2021. doi:10.1038/s41573-020-00117-w

    doi:10.1038/s41573-020-00117-w 2021

  8. [8]

    Chattopadhay, A., Sarkar, A., Howlader, P., and Balasubramanian, V. N. Grad-cam++: Improved visual explanations for deep convolutional networks. In 2018 IEEE Winter Conference on Applications of Computer Vision (WACV), pp.\ 839--847. IEEE, 2018

    work page 2018

Show all 56 references
  1. [9]

    A simple framework for contrastive learning of visual representations

    Chen, T., Kornblith, S., Norouzi, M., and Hinton, G. A simple framework for contrastive learning of visual representations. In International Conference on Machine Learning, pp.\ 1597--1607, 2020

    2020

  2. [10]

    and Yanardag, P

    Dalva, Y. and Yanardag, P. Noiseclr: A contrastive learning approach for unsupervised discovery of interpretable directions in diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp.\ 24209--24218, June 2024

    2024

  3. [11]

    and Nichol, A

    Dhariwal, P. and Nichol, A. Diffusion models beat gans on image synthesis, 2021

    2021

  4. [12]

    An image is worth 16x16 words: Transformers for image recognition at scale

    Dosovitskiy, A., Beyer, L., Kolesnikov, A., Weissenborn, D., Zhai, X., Unterthiner, T., Dehghani, M., Minderer, M., Heigold, G., Gelly, S., Uszkoreit, J., and Houlsby, N. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020

    2010 arXiv

  5. [13]

    et al, A. C. Cellprofiler: Image analysis software for identifying and quantifying cell phenotypes. Genome biology, 7: 0 R100, 02 2006. doi:10.1186/gb-2006-7-10-r100

    2006 doi

  6. [14]

    et al., A. L. Revealing invisible cell phenotypes with conditional generative modeling. Nature Communications, 14, 2022. URL https://api.semanticscholar.org/CorpusID:249873188

    2022

  7. [15]

    et al., P. D. C. High-Content Phenotypic Profiling of Drug Response Signatures across Distinct Cancer Cells . Molecular Cancer Therapeutics, 9 0 (6): 0 1913--1926, 06 2010. ISSN 1535-7163. doi:10.1158/1535-7163.MCT-09-1148. URL https://doi.org/10.1158/1535-7163.MCT-09-1148

    1913 doi

  8. [16]

    Cert: Contrastive self-supervised learning for language understanding

    Fang, M., Smith, A., Guo, H., et al. Cert: Contrastive self-supervised learning for language understanding. arXiv preprint arXiv:2005.12766, 2020

    2005 arXiv

  9. [17]

    Simcse: Simple contrastive learning of sentence embeddings

    Gao, T., Yao, X., and Chen, D. Simcse: Simple contrastive learning of sentence embeddings. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pp.\ 6894--6910, 2021

    2021

  10. [18]

    Ganalyze: Toward visual definitions of cognitive image properties

    Goetschalckx, L., Andonian, A., Oliva, A., and Isola, P. Ganalyze: Toward visual definitions of cognitive image properties. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp.\ 5743--5752, October 2019. doi:10.1109/ICCV.2019.00584

    2019

  11. [19]

    J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., and Bengio, Y

    Goodfellow, I. J., Pouget-Abadie, J., Mirza, M., Xu, B., Warde-Farley, D., Ozair, S., Courville, A., and Bengio, Y. Generative adversarial networks, 2014

    2014

  12. [20]

    Diffusion models in bioinformatics: A new wave of deep learning revolution in action

    Guo, Z., Liu, J., Wang, Y., Chen, M., Wang, D., Xu, D., and Cheng, J. Diffusion models in bioinformatics: A new wave of deep learning revolution in action. CoRR, abs/2302.10907, 2023. URL https://arxiv.org/abs/2302.10907

    2023 arXiv

  13. [21]

    Deep residual learning for image recognition

    He, K., Zhang, X., Ren, S., and Sun, J. Deep residual learning for image recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR), pp.\ 770--778, 2016

    2016

  14. [22]

    Denoising diffusion probabilistic models, 2020

    Ho, J., Jain, A., and Abbeel, P. Denoising diffusion probabilistic models, 2020

    2020

  15. [23]

    Huang, G., Liu, Z., Van Der Maaten, L., and Weinberger, K. Q. Densely connected convolutional networks. In Proceedings of the IEEE conference on computer vision and pattern recognition (CVPR), pp.\ 4700--4708, 2017

    2017

  16. [24]

    Adversarial counterfactual visual explanations

    Jeanneret, G., Simon, L., and Jurie, F. Adversarial counterfactual visual explanations. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp.\ 16425--16435, June 2023

    2023

  17. [25]

    Text-to-image models for counterfactual explanations: A black-box approach

    Jeanneret, G., Simon, L., and Jurie, F. Text-to-image models for counterfactual explanations: A black-box approach. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pp.\ 4757--4767, January 2024

    2024

  18. [26]

    and Park, J

    Kang, M. and Park, J. Contragan: Contrastive learning for conditional image generation. In Advances in Neural Information Processing Systems, volume 33, pp.\ 21312--21323, 2020

    2020

  19. [27]

    A style-based generator architecture for generative adversarial networks

    Karras, T., Laine, S., and Aila, T. A style-based generator architecture for generative adversarial networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp.\ 4401--4410, 2019

    2019

  20. [28]

    Analyzing and improving the image quality of StyleGAN

    Karras, T., Laine, S., Aittala, M., Hellsten, J., Lehtinen, J., and Aila, T. Analyzing and improving the image quality of StyleGAN . In Proc. CVPR, 2020

    2020

  21. [29]

    Kingma, D. P. and Welling, M. Auto-encoding variational bayes. Proceedings of the International Conference on Learning Representations (ICLR), 2014. URL https://arxiv.org/abs/1312.6114

    2014 arXiv

  22. [30]

    Diffusion models already have a semantic latent space, 2023

    Kwon, M., Jeong, J., and Uh, Y. Diffusion models already have a semantic latent space, 2023. URL https://arxiv.org/abs/2210.10960

    2023 arXiv

  23. [31]

    T., Isola, P., Globerson, A., Irani, M., and Mosseri, I

    Lang, O., Gandelsman, Y., Yarom, M., Wald, Y., Elidan, G., Hassidim, A., Freeman, W. T., Isola, P., Globerson, A., Irani, M., and Mosseri, I. Explaining in style: Training a gan to explain a classifier in stylespace. In Proceedings of the IEEE/CVF International Conference on C...

    2021

  24. [32]

    T., Isola, P., Globerson, A., Irani, M., and Mosseri, I

    Lang, O., Gandelsman, Y., Yarom, M., Wald, Y., Elidan, G., Hassidim, A., Freeman, W. T., Isola, P., Globerson, A., Irani, M., and Mosseri, I. Explaining in style: Training a gan to explain a classifier in stylespace. arXiv preprint arXiv:2104.13369, 2021 b

    2021 arXiv

  25. [33]

    A survey of convolutional neural networks: Analysis, applications, and prospects, 2020

    Li, Z., Yang, W., Peng, S., and Liu, F. A survey of convolutional neural networks: Analysis, applications, and prospects, 2020

    2020

  26. [34]

    A convnet for the 2020s

    Liu, Z., Mao, H., Wu, C.-Y., Feichtenhofer, C., Darrell, T., and Xie, S. A convnet for the 2020s. arXiv preprint arXiv:2201.03545, 2022

    2022 arXiv

  27. [35]

    L., Srivatsan, S

    Lotfollahi, M., Klimovskaia Susmelj, A., De Donno, C., Hetzel, L., Ji, Y., Ibarra, I. L., Srivatsan, S. R., Naghipourfar, M., Daza, R. M., Martin, B., Shendure, J., McFaline-Figueroa, J. L., Boyeau, P., Wolf, F. A., Yakubova, N., Günnemann, S., Trapnell, C., Lopez-Paz, D., and...

    2023 doi

  28. [36]

    A bird’s-eye view of deep learning in bioimage analysis

    Meijering, E. A bird’s-eye view of deep learning in bioimage analysis. Computational and Structural Biotechnology Journal, 18: 0 2312--2325, 2020. ISSN 2001-0370. doi:https://doi.org/10.1016/j.csbj.2020.08.003. URL https://www.sciencedirect.com/science/article/pii/S2001037020303561

    2020 doi

  29. [37]

    A., Carpenter, A

    Moshkov, N., Bornholdt, M., Benoit, S., Smith, M., McQuin, C., Goodman, A., Senft, R., Han, Y., Babadi, M., Horvath, P., Cimini, B. A., Carpenter, A. E., Singh, S., and Caicedo, J. C. Learning representations for image-based profiling of perturbations. bioRxiv, 2022. doi:10.11...

    2022 doi

  30. [38]

    Nichol, A. Q. and Dhariwal, P. Improved denoising diffusion probabilistic models. In Meila, M. and Zhang, T. (eds.), Proceedings of the 38th International Conference on Machine Learning, volume 139 of Proceedings of Machine Learning Research, pp.\ 8162--8171. PMLR, 18--24 Jul ...

    2021

  31. [40]

    W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al

    Radford, A., Kim, J. W., Hallacy, C., Ramesh, A., Goh, G., Agarwal, S., Sastry, G., Askell, A., Mishkin, P., Clark, J., et al. Learning transferable visual models from natural language supervision. In International Conference on Machine Learning, 2021

    2021

  32. [42]

    High-resolution image synthesis with latent diffusion models

    Rombach, R., Blattmann, A., Lorenz, D., Esser, P., and Ommer, B. High-resolution image synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp.\ 10684--10695, June 2022 b . doi:10.1109/CVPR52688.20...

    2022

  33. [43]

    U-net: Convolutional networks for biomedical image segmentation

    Ronneberger, O., Fischer, P., and Brox, T. U-net: Convolutional networks for biomedical image segmentation. CoRR, abs/1505.04597, 2015. URL http://arxiv.org/abs/1505.04597

    2015 arXiv

  34. [44]

    Facenet: A unified embedding for face recognition and clustering

    Schroff, F., Kalenichenko, D., and Philbin, J. Facenet: A unified embedding for face recognition and clustering. In Proceedings of the IEEE conference on computer vision and pattern recognition, pp.\ 815--823, 2015

    2015

  35. [45]

    R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., and Batra, D

    Selvaraju, R. R., Cogswell, M., Das, A., Vedantam, R., Parikh, D., and Batra, D. Grad-cam: Visual explanations from deep networks via gradient-based localization. In Proceedings of the IEEE International Conference on Computer Vision (ICCV), pp.\ 618--626, 2017

    2017

  36. [46]

    Explanation by progressive exaggeration

    Singla, S., Pollack, B., Chen, J., and Batmanghelich, K. Explanation by progressive exaggeration. In Proceedings of the International Conference on Learning Representations (ICLR), 2020. URL https://openreview.net/forum?id=r1xDBaEKvH

    2020

  37. [47]

    and Biecek, P

    Sobieski, B. and Biecek, P. Global counterfactual directions. In Proceedings of the European Conference on Computer Vision (ECCV), 2024. doi:10.48550/ARXIV.2404.12488. URL https://arxiv.org/abs/2404.12488

    2024 doi arXiv

  38. [48]

    Denoising diffusion implicit models, 2022

    Song, J., Meng, C., and Ermon, S. Denoising diffusion implicit models, 2022

    2022

  39. [49]

    and Ermon, S

    Song, Y. and Ermon, S. Generative modeling by estimating gradients of the data distribution, 2020

    2020

  40. [50]

    Representation learning with contrastive predictive coding, 2019

    van den Oord, A., Li, Y., and Vinyals, O. Representation learning with contrastive predictive coding, 2019. URL https://arxiv.org/abs/1807.03748

    2019 arXiv

  41. [51]

    and Babenko, A

    Voynov, A. and Babenko, A. Unsupervised discovery of interpretable directions in the GAN latent space. In Proceedings of the 37th International Conference on Machine Learning (ICML), volume 119 of Proceedings of Machine Learning Research, pp.\ 9786--9796. PMLR, 2020. URL https...

    2020

  42. [52]

    and Liu, H

    Wang, F. and Liu, H. Understanding the behaviour of contrastive loss. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pp.\ 2495--2504, 2021

    2021

  43. [53]

    Deep learning in microscopy image analysis: A survey

    Xing, F., Xie, Y., Su, H., Liu, F., and Yang, L. Deep learning in microscopy image analysis: A survey. IEEE Transactions on Neural Networks and Learning Systems, 29 0 (10): 0 4550--4568, 2018. doi:10.1109/TNNLS.2017.2766168

    2018

  44. [54]

    Yang, S., Hwang, H., and Ye, J. C. Zero-shot contrastive loss for text-guided diffusion image style transfer. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2023. URL https://openaccess.thecvf.com/content/ICCV2023/html/Yang_Zero-Shot_Contras...

    2023

  45. [55]

    K., Simsar, E., Er, E

    Y \"u ksel, O. K., Simsar, E., Er, E. G., and Yanardag, P. Latentclr: A contrastive learning approach for unsupervised discovery of interpretable directions. In Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), pp.\ 14263--14272, October 2021

    2021

  46. [56]

    Zeiler, M. D. and Fergus, R. Visualizing and understanding convolutional networks. In Fleet, D., Pajdla, T., Schiele, B., and Tuytelaars, T. (eds.), Computer Vision -- ECCV 2014, pp.\ 818--833, Cham, 2014. Springer International Publishing. ISBN 978-3-319-10590-1

    2014

  47. [57]

    Zhu, J.-Y., Park, T., Isola, P., and Efros, A. A. Unpaired image-to-image translation using cycle-consistent adversarial networks, 2020

    2020

  48. [58]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 gl...

Pith tools

Reviewed August 8, 2026 · model on record in the stance chip above.