Pith. sign in

REVIEW 3 major objections 8 minor 1 cited by

Dataset Distillation via Vision-Language Category Prototype

T0 review · 3 major / 8 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Adding LLM-written text prototypes to image prototypes improves dataset distillation enough to set reported accuracy highs on six benchmarks.

desk verdict Solid DD results with LLM captions, but the key ablation is confounded by prompt distribution and hyperparameters are tuned on a reported test set. read the letter →

arxiv 2506.23580 v1 pith:7D4F45OP submitted 2025-06-30 cs.CV

Yawen Zou , Guang Li , Duo Su , Zi Wang , Jun Yu , Chao Zhang This is my paper
classification cs.CV
keywords datasetdistillationtextprototypesimagevision-languagediffusionmodelsLLM-generatedcaptionsk-meansclusteringsyntheticdatageneration
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 claims that dataset distillation has been missing semantic information: prior diffusion-based methods compress only what can be seen in image features, so distilled images can lose target objects, warp anatomy, or lock in background co-occurrence biases. The proposed remedy is to give each distilled-image cluster a text prototype extracted from LLM-generated captions, and to condition the diffusion model on both the image prototype and that text prototype. The result, the paper reports, is surrogate datasets that are more logically coherent and that achieve the best top-1 accuracy it compares against on ImageWoof, ImageNette, ImageIDC, ImageNet-1K, CIFAR-10, and CIFAR-100. A sympathetic reader would care because the recipe adds semantic grounding to a purely visual compression task using only open-source components, and it works on datasets that have no pre-existing text annotations.

What carries the argument

The load-bearing object is a cross-modal prototype pair. Image prototypes are cluster centers produced by k-means on latent features after Local Outlier Factor removal of low-density samples. Text prototypes are built per cluster by tokenizing LLM-generated captions, deleting words that appear in more than a threshold fraction of the class, ranking the surviving words by frequency, and selecting the sentence whose matching score, computed by summing the frequencies of its top-k words, is highest. The image prototype and the text prototype are then fed together into a fine-tuned latent diffusion model, with the text embedding conditioning the denoising step, so that every synthesized image inherits both the cluster's visual center and its semantic description.

What would settle it

Re-run the pipeline with the text prototype for each cluster replaced by a randomly chosen caption from a different cluster, keeping all image prototypes and diffusion settings fixed; if top-1 accuracy on ImageNette at IPC 50 does not fall materially below the reported 81.2%, then the word-frequency text selection is not what carries the claimed gain. A complementary check is to compute a standard vision-language embedding similarity between each text prototype and its own cluster's images: if prototypes are no closer to their own cluster than to other clusters, the semantic-matching mechanism fails.

Watch

Extended reading notes

Core claim

The central discovery is that the missing ingredient in diffusion-based dataset distillation is semantic, not visual. Once each image prototype, obtained by k-means clustering in the latent space of a pretrained autoencoder, is paired with a text prototype selected from LLM-generated descriptions by word-frequency matching, the latent diffusion model produces synthetic images that contain the intended objects, avoid structurally implausible outputs, and train downstream classifiers to higher top-1 accuracy than image-only distillation on every benchmark tested. Concretely, the paper reports gains of 3.9%, 4.9%, and 3.5% over the best prior method on ImageNette and 2.9%, 4.2%, and 2.5% on ImageIDC at IPC 10, 20, and 50, plus an 8.0% gain over RDED on CIFAR-100 at IPC 10. The claim is that this is the first integration of language information into visual dataset distillation for classification, and that it generalizes across low- and high-resolution datasets and across test architectures.

Load-bearing premise

The method assumes the LLM-generated captions are accurate and consistent enough that the sentence picked by word-frequency matching truly describes what the cluster's images show, but caption quality is never measured directly, only the final downstream classifier accuracy.

Editorial extensions

If this is right

  • Datasets without any existing text annotations can now be distilled with semantic guidance, because the LLM supplies the captions in a one-off pass.
  • The paired text prototype fixes two failure modes of earlier diffusion-based distillation: synthesized images that contain no target object and synthesized images with implausible structures such as extra or missing limbs.
  • Per-cluster text prototypes split a class into distinct visual contexts, so the distilled dataset preserves within-class diversity rather than collapsing to a single class-level description.
  • The gains are largest at small images-per-class settings, for example 8.0% on CIFAR-100 at IPC 10, suggesting semantic information is most valuable when the number of synthesized images is tight.
  • Because the whole pipeline uses only public datasets, an open-source LLM, and a public diffusion model, the reported results establish a reusable baseline for future vision-language dataset distillation.

Reading between the lines

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

  • The mechanism suggests text prototypes could be edited directly to steer distilled data, for example rewriting captions to remove co-occurrence biases such as fish always appearing with plants; the paper does not explore this editing use.
  • Word-frequency matching is one proxy for semantic closeness; a direct test would replace Equation 4 with embedding-similarity selection of the caption per cluster, and if that matches or beats the proposed method, the frequency ranking is serving as a proxy rather than the essential ingredient.
  • The same prototype-plus-caption scheme may extend beyond classification to detection or segmentation distillation by prompting the LLM for layout-aware descriptions, which the paper only lists as future work.
  • Because the text stage is a single LLM pass, the two-stage pipeline likely keeps caption-generation cost low relative to iterative pixel optimization on very large source sets, making the semantic addition cheap at scale.
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 / 8 minor

Summary. This paper proposes a dataset distillation framework that augments image prototypes with text prototypes. The authors use LLaVA to generate per-image descriptions, fine-tune a Stable Diffusion (LDM) model on the resulting image-text pairs, cluster the latent features per class with k-means, and select, for each cluster, a 'text prototype' by ranking cluster-specific words by frequency and choosing the sentence with the highest matching score (Eq. 4). The cluster center (image prototype) and the text prototype jointly condition the LDM to synthesize distilled images. Experiments on ImageWoof, ImageNette, ImageIDC, ImageNet-1K, CIFAR-10, and CIFAR-100 report consistent improvements over prior methods across IPC settings and architectures, with the largest gains on ImageNette (e.g., +4.9 points at IPC-20) and on CIFAR-100 (+8.0 points at IPC-10 over RDED).

Significance. The method is a timely extension of generative dataset distillation: it exploits LLM-generated captions for datasets that have no text annotations, and the released code and generated data support reproducibility. If the mechanism claims are confirmed, the frequency-based text prototype selection would be a useful, compute-light component for diffusion-based dataset distillation. The empirical picture is broad (six datasets, three architectures, IPC from 10 to 100), and the pipeline-level gains over strong baselines such as Minimax and D4M are consistent. The main weaknesses are that the component-level ablation (Table 5) does not isolate the proposed selection rule from a prompt-distribution confound, and the parameter analysis in Section 4.6 selects hyperparameters on the ImageIDC benchmark, which is itself reported as a headline result.

major comments (3)
  1. [§4.5, Table 5] The ablation that isolates the proposed text-prototype selection (DCS vs. GGS and L+FK) is confounded by prompt distribution. The LDM is fine-tuned on LLaVA-generated captions (§3.1), and DCS text prototypes are drawn from that same caption set, whereas §4.5 states that GGS sentences are 'based on FK not encountered during the model's training'; L+FK likewise feeds the model keyword bags rather than the sentence distribution used in fine-tuning. The observed DCS advantage could therefore reflect distribution match rather than the word-frequency matching in Eq. 4. The authors should add the missing control of a text prototype selected at random from the same cluster's LLaVA captions (in-distribution but without frequency-based selection), and ideally fine-tune an LDM whose training corpus also includes GGS-style sentences so that the GGS baseline is in-distribution. Without these controls, the specific contribution of the Eq. 4 selection rule over simply using any LLaVA caption is not established.
  2. [§4.6, Fig. 4] Hyperparameters α (LOF contamination), β (nonrepresentative threshold), and k (top-k words) are analyzed on ImageIDC, and the text indicates that β=0.2 and k=35 are selected from those curves while α is 'adjusted based on the dataset characteristics.' The paper never states which values are actually used in Tables 1–4. Because ImageIDC is itself one of the reported benchmarks, choosing β and k to maximize ImageIDC accuracy and then reporting ImageIDC as a primary result is a form of test-set hyperparameter selection that can inflate the headline numbers. The authors should state the exact β, k, and α settings used in every experiment, and either select them on a held-out split or show that the same settings perform well across all datasets reported.
  3. [Tables 3–4] The state-of-the-art claim for the low-resolution benchmarks is not fully supported by the comparison set: CIFAR-10 and CIFAR-100 are evaluated only against SRe2L and RDED, omitting standard baselines such as DM, IDC, MTT, and DSA that report on these datasets, and ImageNet-1K (Table 3) omits generative baselines such as GLaD that report the same benchmark. The comparison should be expanded to at least the strongest available baselines for each dataset, or the SOTA claim should be explicitly qualified to the compared methods.
minor comments (8)
  1. [§3.4, Eq. (6)] The notation 'Concat(zc_t, τθ(Tc))' is not how Stable Diffusion conditions on text (which is via cross-attention), so Eq. (6) and Algorithm 1 should be corrected or clarified to avoid implying channel-wise concatenation of text embeddings with latents.
  2. [§3.3.2, Eq. (3)] Equation (3) defines Rw as a set of words, but Eq. (4) uses Rw[w] as an importance weight; the authors should clarify that Rw is a (word, frequency) set and that Rw[w] denotes the cluster-level frequency fc.
  3. [Table 1] The 'Full' column in Table 1 is not a competing method and should be labeled as the full-dataset reference accuracy, with a note in the caption so the bold/underlined convention is unambiguous.
  4. [Contributions, §1] The claim of being the 'first work that integrates language information into visual dataset distillation' is too broad given that GLaD [4] already conditions generative dataset distillation on language; the authors should qualify the novelty to the use of LLM-generated descriptive text prototypes with frequency-based selection.
  5. [§3.1] The paper does not specify which LLaVA checkpoint (e.g., LLaVA-1.5-7B/13B or LLaVA-NeXT) or inference settings produced the captions, nor the exact contamination values per dataset mentioned in §3.2; these details are needed for reproducibility.
  6. [§4.2] The total computational cost of the pipeline (LLaVA captioning, LDM fine-tuning for 8 epochs, and generation) is not reported or compared with the baselines, which matters because dataset distillation is motivated by efficiency.
  7. [Table 2] Some reported gains are within one standard deviation of the baseline (e.g., ImageNette IPC-10: 64.8±3.6 vs. 60.9±1.7); a sentence discussing run-to-run variability would strengthen the SOTA claims, though the consistency across IPC settings and architectures is reassuring.
  8. [References] Reference [8] has a garbled author list ('Zongxion Geng, Jiahui andg Chen...'), and the paper inconsistently typesets 'D4M/D 4M' and 'SRe2L/SRe 2L'; these should be cleaned up.

Circularity Check

1 steps flagged · score 6.0 of 10

ImageIDC headline numbers are selected by tuning alpha, beta, and k on ImageIDC itself.

  1. fitted input called prediction [Section 4.6 (Parameter Analysis) and Section 4.3 / Table 2 (ImageIDC results)]
    "We analyze the sensitivity of parameters α (Contamination), β (Nonrepresentative Threshold), and k (Top-k words) on ImageIDC, as shown in Fig. 4 (a)-(c). ... ImageNette and ImageIDC We assess our method using the ResNetAP-10 architecture under IPC 10, 20, and 50, as shown in Table 2."

    The 'Ours' entries for ImageIDC in Table 2 (e.g., 57.0/63.3/71.9 at IPC 10/20/50) are produced with hyperparameters alpha, beta, and k chosen by peaking the accuracy curves in Fig. 4 on the same ImageIDC dataset. The abstract then headlines '2.9%, 4.2%, and 2.5% on ImageIDC' as a SOTA improvement. Reporting the best configuration found by searching on ImageIDC as a result on ImageIDC means that benchmark's outcome is a fitted maximum, not an independent prediction. Other datasets (ImageWoof, ImageNette, ImageNet-1K, CIFAR) are not used for this tuning, so the circularity is partial.

full rationale

The core pipeline is not definitionally circular: LLaVA captions are generated once, the diffusion model is fine-tuned on them, text prototypes are selected by word-frequency scoring, and final accuracies are measured on held-out test sets. The text prototype is not defined in terms of the final accuracy, so the main derivation chain does not reduce to its inputs. The DCS-vs-GGS ablation has a distribution confound, since DCS prompts are drawn from the LLaVA captions used to fine-tune the LDM while GGS prompts are out-of-distribution; the paper itself notes GGS sentences are 'based on FK not encountered during the model's training.' That is a fairness/correctness issue rather than an equation-level circularity. The one clear circular step is the ImageIDC hyperparameter selection: Section 4.6 tunes alpha, beta, and k by inspecting accuracy on ImageIDC, and the same dataset is then reported as a headline SOTA benchmark. Those ImageIDC numbers are therefore the result of fitting to that dataset's test accuracy, not an independent prediction. Other benchmarks remain independent, so the circularity is partial and confined to ImageIDC.

Assumptions & free parameters 3 free parameters · 3 assumptions · 0 invented entities

The method introduces no new physical or conceptual entities. It uses existing components (LLaVA, Stable Diffusion, K-means, LOF) plus a heuristic for selecting text prototypes. The main free parameters are the LOF contamination, the nonrepresentative threshold, and the top-k word count, all tuned empirically.

free parameters (3)
  • alpha (LOF contamination) = not stated in main text
    LOF contamination parameter in outlier removal; Section 4.6 shows sensitivity, but the final value used for main results is not given.
  • beta (nonrepresentative threshold) = 0.2 (implied by text)
    Words appearing in more than 20% of class samples are excluded; selected based on ImageIDC performance in Section 4.6.
  • k (top-k words) = not stated in main text
    Number of representative words used in Eq. 3; tuned in Section 4.6 with a maximum at 35 for lower IPC, but final value not specified.
assumptions (3)
  • domain assumption LLaVA-generated descriptions are semantically accurate and sufficiently diverse to serve as text prototypes.
    The method relies on these captions for both fine-tuning the diffusion model and selecting text prototypes; incorrect or noisy captions could mislead generation. See Section 3.1.
  • ad hoc to paper The word-frequency-based text prototype selection identifies the sentence most representative of each cluster.
    The scoring in Eq. 4 is a heuristic with no theoretical justification; its success is empirical. See Section 3.3.2.
  • standard math Standard diffusion model and K-means assumptions (e.g., that cluster centers capture visual modes).
    Background methods from prior work; no novel mathematical claim.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Dataset Distillation via Vision-Language Category Prototype." pith.science (2026). https://pith.science/paper/7D4F45OP

@misc{pith2026250623580,
  author       = {Pith},
  title        = {Pith review of: Dataset Distillation via Vision-Language Category Prototype},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7D4F45OP}},
  note         = {Machine review of arXiv:2506.23580}
}
read the original abstract

Dataset distillation (DD) condenses large datasets into compact yet informative substitutes, preserving performance comparable to the original dataset while reducing storage, transmission costs, and computational consumption. However, previous DD methods mainly focus on distilling information from images, often overlooking the semantic information inherent in the data. The disregard for context hinders the model's generalization ability, particularly in tasks involving complex datasets, which may result in illogical outputs or the omission of critical objects. In this study, we integrate vision-language methods into DD by introducing text prototypes to distill language information and collaboratively synthesize data with image prototypes, thereby enhancing dataset distillation performance. Notably, the text prototypes utilized in this study are derived from descriptive text information generated by an open-source large language model. This framework demonstrates broad applicability across datasets without pre-existing text descriptions, expanding the potential of dataset distillation beyond traditional image-based approaches. Compared to other methods, the proposed approach generates logically coherent images containing target objects, achieving state-of-the-art validation performance and demonstrating robust generalization. Source code and generated data are available in https://github.com/zou-yawen/Dataset-Distillation-via-Vision-Language-Category-Prototype/

Figures

Figures reproduced from arXiv: 2506.23580 by the authors.

Figure 1
Figure 1. Visualization results of SRe2L, GR (w/o text), and GR (w/ text). GR (w/o or w/ text) denotes the generative model outputs without and with text descriptions. Notably, GR (w/ text) captures rich details of target objects while preserving background diver￾sity, leading to more comprehensive and visually coherent images. surrogate dataset helps alleviate concerns about privacy and copyright issues [2, 7, 15]. Dataset d… view at source ↗
Figure 2
Figure 2. Overview of the proposed framework. The framework starts with generating image-text pairs using the LLaVA model, followed [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Visualization of images generated using different semantic strategies. For each column, the images are generated using the same [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Parameter Analysis of α (Contamination), β (Nonrepresentative Threshold), and k (Top-k words) on ImageIDC. trend, other settings reach a peak at 0.2 before declining and stabilizing. Words appearing in more than 20% of the sam￾ples in each class are classified as nonre…

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Hard Labels In! Rethinking the Role of Hard Labels in Mitigating Local Semantic Drift

    cs.CV 2025-12 conditional novelty 6.0 of 10

    A soft-hard-soft training schedule uses hard labels as an intermediate anchor to correct local semantic drift and improves accuracy under 100x-reduced soft-label storage.

Reference graph

Works this paper leans on

48 extracted references · 40 canonical work pages · cited by 1 Pith paper

  1. [1]

    A review of local outlier factor algorithms for out- lier detection in big data streams

    Omar Alghushairy, Raed Alsini, Terence Soule, and Xiao- gang Ma. A review of local outlier factor algorithms for out- lier detection in big data streams. Big Data and Cognitive Computing, 5(1):1, 2020. 3

    work page 2020

  2. [2]

    No Free Lunch in "Privacy for Free: How does Dataset Condensation Help Privacy"

    Nicholas Carlini, Vitaly Feldman, and Milad Nasr. No free lunch in” privacy for free: How does dataset condensation help privacy”. arXiv preprint arXiv:2209.14987, 2022. 1

    work page Pith review arXiv 2022

  3. [3]

    Dataset distillation by matching training trajectories

    George Cazenavette, Tongzhou Wang, Antonio Torralba, Alexei A Efros, and Jun-Yan Zhu. Dataset distillation by matching training trajectories. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 4750–4759, 2022. 1

    work page 2022

  4. [4]

    Generalizing dataset distillation via deep generative prior

    George Cazenavette, Tongzhou Wang, Antonio Torralba, Alexei A Efros, and Jun-Yan Zhu. Generalizing dataset distillation via deep generative prior. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 3739–3748, 2023. 5

    work page 2023

  5. [5]

    Imagenet: A large-scale hierarchical im- age database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical im- age database. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 248–255, 2009. 1, 4

    work page 2009

  6. [6]

    Remember the past: Distilling datasets into addressable memories for neural net- works

    Zhiwei Deng and Olga Russakovsky. Remember the past: Distilling datasets into addressable memories for neural net- works. In Proceedings of the Advances in Neural Infor- mation Processing Systems (NeurIPS), pages 34391–34404,

    work page

  7. [7]

    Privacy for free: How does dataset condensation help privacy? In Proceed- ings of the International Conference on Machine Learning (ICML), pages 5378–5396, 2022

    Tian Dong, Bo Zhao, and Lingjuan Lyu. Privacy for free: How does dataset condensation help privacy? In Proceed- ings of the International Conference on Machine Learning (ICML), pages 5378–5396, 2022. 1

    work page 2022

  8. [8]

    A survey on dataset distillation: Approaches, applications and future directions

    Zongxion Geng, Jiahui andg Chen, Yuandou Wang, Herbert Woisetschlaeger, Sonja Schimmler, Ruben Mayer, Zhiming Zhao, and Chunming Rong. A survey on dataset distillation: Approaches, applications and future directions. In Proceed- ings of the International Joint Conference on Artificial Intel- ligence (IJCAI), 2023. 1

    work page 2023

Show all 48 references
  1. [9]

    Efficient dataset distillation via minimax diffusion

    Jianyang Gu, Saeed Vahidian, Vyacheslav Kungurtsev, Hao- nan Wang, Wei Jiang, Yang You, and Yiran Chen. Efficient dataset distillation via minimax diffusion. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 15793–15803, 2024. 1...

    2024

  2. [10]

    A smaller subset of 10 easily classified classes from imagenet, and a little more french, 2020

    Jeremy Howard. A smaller subset of 10 easily classified classes from imagenet, and a little more french, 2020. URL https://github. com/fastai/imagenette. 2, 4

    2020

  3. [11]

    Dataset condensation via efficient synthetic- data parameterization

    Jang-Hyun Kim, Jinuk Kim, Seong Joon Oh, Sangdoo Yun, Hwanjun Song, Joonhyun Jeong, Jung-Woo Ha, and Hyun Oh Song. Dataset condensation via efficient synthetic- data parameterization. In Proceedings of the International Conference on Machine Learning (ICML) , pages 11102– 1111...

    2022

  4. [12]

    Learning multiple layers of features from tiny images

    Alex Krizhevsky, Geoffrey Hinton, et al. Learning multiple layers of features from tiny images. 2009. 4

    2009

  5. [13]

    Deep learning

    Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. Deep learning. nature, 521(7553):436–444, 2015. 1

    2015

  6. [14]

    Dataset condensation with con- trastive signals

    Saehyung Lee, Sanghyuk Chun, Sangwon Jung, Sangdoo Yun, and Sungroh Yoon. Dataset condensation with con- trastive signals. In Proceedings of the International Con- ference on Machine Learning (ICML), pages 12352–12364,

  7. [15]

    A comprehensive survey to dataset distillation

    Shiye Lei and Dacheng Tao. A comprehensive survey to dataset distillation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 46(1):17–32, 2023. 1

    2023

  8. [16]

    Awesome dataset distillation

    Guang Li, Bo Zhao, and Tongzhou Wang. Awesome dataset distillation. https : / / github . com / Guang000 / Awesome-Dataset-Distillation, 2022. 1

    2022

  9. [17]

    Visual instruction tuning

    Haotian Liu, Chunyuan Li, Qingyang Wu, and Yong Jae Lee. Visual instruction tuning. In Proceedings of the Advances in Neural Information Processing Systems (NeurIPS), pages 34892–34916, 2023. 3

    2023

  10. [18]

    Improved baselines with visual instruction tuning

    Haotian Liu, Chunyuan Li, Yuheng Li, and Yong Jae Lee. Improved baselines with visual instruction tuning. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 26296–26306, 2024

    2024

  11. [19]

    Llava-next: Im- proved reasoning, ocr, and world knowledge

    Haotian Liu, Chunyuan Li, Yuheng Li, Bo Li, Yuanhan Zhang, Sheng Shen, and Yong Jae Lee. Llava-next: Im- proved reasoning, ocr, and world knowledge. https: / / llava - vl . github . io / blog / 2024 - 01 - 30 - llava-next/, 2024. 3

    2024

  12. [20]

    Graph condensation via receptive field distribution matching

    Mengyang Liu, Shanchuan Li, Xinshi Chen, and Le Song. Graph condensation via receptive field distribution matching. arXiv preprint arXiv:2206.13697, 2022. 1

    2022 arXiv

  13. [21]

    The evolution of dataset distillation: Toward scalable and generalizable solutions

    Ping Liu and Jiawei Du. The evolution of dataset distillation: Toward scalable and generalizable solutions. arXiv preprint arXiv:2502.05673, 2025. 1

    2025 arXiv

  14. [22]

    Dream: Efficient dataset distillation by rep- resentative matching

    Yanqing Liu, Jianyang Gu, Kai Wang, Zheng Zhu, Wei Jiang, and Yang You. Dream: Efficient dataset distillation by rep- resentative matching. In Proceedings of the IEEE/CVF In- ternational Conference on Computer Vision (ICCV) , pages 17314–17324, 2023. 1

    2023

  15. [23]

    Efficient dataset distillation using random feature ap- proximation

    Noel Loo, Ramin Hasani, Alexander Amini, and Daniela Rus. Efficient dataset distillation using random feature ap- proximation. In Proceedings of the Advances in Neural Information Processing Systems (NeurIPS) , pages 13877– 13891, 2022. 1

    2022

  16. [24]

    Dataset distillation with convexified implicit gradients

    Noel Loo, Ramin Hasani, Mathias Lechner, and Daniela Rus. Dataset distillation with convexified implicit gradients. In Proceedings of the International Conference on Machine Learning (ICML), pages 22649–22674, 2023. 1

    2023

  17. [25]

    Latent dataset distillation with diffusion models

    Brian B Moser, Federico Raue, Sebastian Palacio, Stanislav Frolov, and Andreas Dengel. Latent dataset distillation with diffusion models. arXiv preprint arXiv:2403.03881, 2024. 1

    2024

  18. [26]

    Dataset meta-learning from kernel ridge-regression

    Timothy Nguyen, Zhourong Chen, and Jaehoon Lee. Dataset meta-learning from kernel ridge-regression. In Proceedings of the International Conference on Learning Representations (ICLR), 2021. 1

    2021

  19. [27]

    Scalable diffusion models with transformers

    William Peebles and Saining Xie. Scalable diffusion models with transformers. In Proceedings of the IEEE/CVF interna- tional conference on computer vision (CVPR) , pages 4195– 4205, 2023. 5, 6

    2023

  20. [28]

    High-resolution image 9 synthesis with latent diffusion models

    Robin Rombach, Andreas Blattmann, Dominik Lorenz, Patrick Esser, and Bj ¨orn Ommer. High-resolution image 9 synthesis with latent diffusion models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 10684–10695, 2022. 2

    2022

  21. [29]

    Data distillation: A survey

    Noveen Sachdeva and Julian McAuley. Data distillation: A survey. Transactions on Machine Learning Research, 2023. 1

    2023

  22. [30]

    Active learning for convolu- tional neural networks: A core-set approach

    Ozan Sener and Silvio Savarese. Active learning for convolu- tional neural networks: A core-set approach. In Proceedings of the International Conference on Learning Representations (ICLR), 2018. 5

    2018

  23. [31]

    Dˆ 4: Dataset distillation via disentangled diffusion model

    Duo Su, Junjie Hou, Weizhi Gao, Yingjie Tian, and Bowen Tang. Dˆ 4: Dataset distillation via disentangled diffusion model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) , pages 5809–5818, 2024. 1, 2, 3, 5

    2024

  24. [32]

    Soft-label dataset distillation and text dataset distillation

    Ilia Sucholutsky and Matthias Schonlau. Soft-label dataset distillation and text dataset distillation. In Proceedings of the International Joint Conference on Neural Networks (IJCNN), pages 1–8, 2021. 1

    2021

  25. [33]

    On the diversity and realism of distilled dataset: An efficient dataset distilla- tion paradigm

    Peng Sun, Bei Shi, Daiwei Yu, and Tao Lin. On the diversity and realism of distilled dataset: An efficient dataset distilla- tion paradigm. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 9390–9399, 2024. 5, 6

    2024

  26. [34]

    Con- trastive multiview coding

    Yonglong Tian, Dilip Krishnan, and Phillip Isola. Con- trastive multiview coding. In Proceedings of the European Conference on Computer Vision (ECCV) , pages 776–794,

  27. [35]

    Diffusers: State-of-the-art diffu- sion models

    Patrick von Platen, Suraj Patil, Anton Lozhkov, Pedro Cuenca, Nathan Lambert, Kashif Rasul, Mishig Davaadorj, Dhruv Nair, Sayak Paul, William Berman, Yiyi Xu, Steven Liu, and Thomas Wolf. Diffusers: State-of-the-art diffu- sion models. https://github.com/huggingface/ diffusers...

    2022

  28. [36]

    Cafe: Learning to condense dataset by align- ing features

    Kai Wang, Bo Zhao, Xiangyu Peng, Zheng Zhu, Shuo Yang, Shuo Wang, Guan Huang, Hakan Bilen, Xinchao Wang, and Yang You. Cafe: Learning to condense dataset by align- ing features. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages ...

    2022

  29. [37]

    DiM: Distilling dataset into genera- tive model

    Kai Wang, Jianyang Gu, Daquan Zhou, Zheng Zhu, Wei Jiang, and Yang You. DiM: Distilling dataset into genera- tive model. arXiv preprint arXiv:2303.04707, 2023. 1

    2023 arXiv

  30. [38]

    Tongzhou Wang, Jun-Yan Zhu, Antonio Torralba, and Alexei A. Efros. Dataset distillation. arXiv preprint arXiv:1811.10959, 2018. 1, 2

    2018 arXiv

  31. [39]

    Herding dynamical weights to learn

    Max Welling. Herding dynamical weights to learn. In Pro- ceedings of the international conference on machine learn- ing (ICML), pages 1121–1128, 2009. 5

    2009

  32. [40]

    Squeeze, recover and relabel: Dataset condensation at imagenet scale from a new perspective

    Zeyuan Yin, Eric Xing, and Zhiqiang Shen. Squeeze, recover and relabel: Dataset condensation at imagenet scale from a new perspective. In Proceedings of the Advances in Neural Information Processing Systems (NeurIPS), 2023. 5

    2023

  33. [41]

    Squeeze, recover and relabel: Dataset condensation at imagenet scale from a new perspective

    Zeyuan Yin, Eric Xing, and Zhiqiang Shen. Squeeze, recover and relabel: Dataset condensation at imagenet scale from a new perspective. In Proceedings of the Advances in Neural Information Processing Systems (NeurIPS), 2024. 6

    2024

  34. [42]

    A compre- hensive survey to dataset distillation

    Ruonan Yu, Songhua Liu, and Xinchao Wang. A compre- hensive survey to dataset distillation. IEEE Transactions on Pattern Analysis and Machine Intelligence , 46(1):150–170,

  35. [43]

    Dataset condensation with dif- ferentiable siamese augmentation

    Bo Zhao and Hakan Bilen. Dataset condensation with dif- ferentiable siamese augmentation. In Proceedings of the In- ternational Conference on Machine Learning (ICML), pages 12674–12685, 2021. 1

    2021

  36. [44]

    Dataset condensation with gra- dient matching

    Bo Zhao and Hakan Bilen. Dataset condensation with gra- dient matching. In Proceedings of the International Confer- ence on Learning Representations (ICLR), 2021. 1, 5

    2021

  37. [45]

    Synthesizing informative training samples with gan

    Bo Zhao and Hakan Bilen. Synthesizing informative training samples with gan. In Proceedings of the Advances in Neu- ral Information Processing Systems (NeurIPS), Workshop ,

  38. [46]

    Dataset condensation with distri- bution matching

    Bo Zhao and Hakan Bilen. Dataset condensation with distri- bution matching. In Proceedings of the IEEE/CVF Confer- ence on Computer Vision and Pattern Recognition (CVPR) , pages 6514–6523, 2023. 1

    2023

  39. [47]

    Im- proved distribution matching for dataset condensation

    Ganlong Zhao, Guanbin Li, Yipeng Qin, and Yizhou Yu. Im- proved distribution matching for dataset condensation. In Proceedings of the IEEE/CVF Conference on Computer Vi- sion and Pattern Recognition (CVPR) , pages 7856–7865,

  40. [48]

    Dataset distillation using neural feature regression

    Yongchao Zhou, Ehsan Nezhadarya, and Jimmy Ba. Dataset distillation using neural feature regression. In Proceedings of the Advances in Neural Information Processing Systems (NeurIPS), pages 9813–9827, 2022. 1 10

    2022

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.