Pith. sign in

REVIEW 4 major objections 3 minor 43 references

BloomCoreset: Fast Coreset Sampling using Bloom Filters for Fine-Grained Self-Supervised Learning

T0 review · 4 major / 3 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read BloomCoreset claims that a Bloom filter built from binarized OpenCLIP features can replace the expensive train-then-sample coreset selection of SimCore, cutting sampling time by 98.5% while keeping average downstream accuracy within 0.83%.

desk verdict The speedup is real, but the Bloom filter as described is a random prefilter; the actual work is top-k on OpenCLIP features. read the letter →

arxiv 2412.16942 v1 pith:SNYDUCR2 submitted 2024-12-22 cs.CV

Prajwal Singh , Gautam Vashishtha , Indra Deep Mastan , Shanmuganathan Raman This is my paper · ORCID
classification cs.CV
keywords coresetsamplingself-supervisedlearningBloomfiltercountingfine-grainedclassificationopen-setOpenCLIPcontrastive
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

This paper tries to establish that coreset sampling for fine-grained self-supervised learning can be made dramatically faster without much loss in quality by using a probabilistic hashing structure instead of training a model on the downstream data first. The proposed method, BloomCoreset, builds a Counting Bloom Filter from binarized OpenCLIP features of the downstream dataset, uses it to quickly filter candidate images from a large unlabeled open-set, then applies a cosine-similarity top-k refinement. Integrated into the SimCore framework, it selects a 1% coreset from ImageNet-1k in about 20 minutes rather than about 23 hours, with an average accuracy trade-off of 0.83% across 11 fine-grained downstream datasets, and it beats SimCore on four of them. This matters because the sampling stage is what makes open-set self-supervised learning impractical as an off-the-shelf tool, and the paper argues the speedup comes with only a small quality penalty.

What carries the argument

The central object is a Counting Bloom Filter (CBF) over binarized OpenCLIP image features. Each downstream image's 512-dimensional feature vector is converted to a binary code by thresholding at zero, and ten murmurhash3 variants set counters in the CBF; open-set images whose binary codes pass the membership test become candidates. Since Bloom filters have false positives but no false negatives, the candidate set is then refined by computing cosine similarity between downstream features and candidates and keeping the top-k, so the exponential speedup of hashing is combined with a cheap exact similarity step on the surviving candidates.

What would settle it

Run BloomCoreset on a downstream fine-grained dataset whose classes are known to be hard for OpenCLIP's zero-shot features, such as car models or dog breeds, and compare the accuracy trade-off against SimCore; if the trade-off is substantially larger than 0.83%, or if the selected coreset has no higher cosine similarity to the downstream data than a random sample, the binarized-feature proxy is the weak point.

Watch

Extended reading notes

Core claim

The central claim is that SimCore's expensive first stage—training a self-supervised encoder on the downstream data to find the closest open-set images—can be replaced by a membership test in a Counting Bloom Filter populated with binarized features from a frozen OpenCLIP encoder, followed by a cosine-similarity top-k filter. With this two-stage pipeline, the paper reports a 98.5% reduction in sampling time and an average accuracy trade-off of 0.83% against SimCore across 11 downstream datasets, with per-dataset trade-offs ranging from −5.4% to +4.48%. The paper also claims the method generalizes across different open-sets (ImageNet-1k, MS COCO, iNaturalist) and that the top-k refinement is necessary because raw Bloom-filter membership alone produces a coreset that does not improve representation learning.

Load-bearing premise

The load-bearing premise is that binarized features from a frozen OpenCLIP encoder preserve enough fine-grained semantic similarity that cosine nearest neighbors in this space are a good proxy for the target-specific neighbors SimCore finds by training on the downstream data.

Editorial extensions

If this is right

  • Coreset sampling for fine-grained SSL can be cut from hours to minutes: on two RTX 3090 GPUs, sampling a 1% ImageNet-1k coreset drops from roughly 23 hours to roughly 20 minutes.
  • A practitioner can swap BloomCoreset into SimCore in place of its sampling stage and expect an average accuracy loss of only 0.83% across 11 datasets, with gains on Action, Indoor, Textures, and Food.
  • The top-k filtering step is load-bearing: without it, Bloom-filter membership alone yields a coreset whose downstream accuracy is much closer to training on the downstream data alone.
  • The speedup transfers to other open-sets, with competitive or better accuracy than SimCore on several downstream targets when MS COCO or iNaturalist is used as the open-set.
  • Because OpenCLIP is frozen and no downstream training is needed for sampling, the bottleneck shifts to feature extraction throughput rather than optimization, so the method scales better as open-sets grow.
  • The 0.83% average trade-off is an average over datasets; the per-dataset range runs from a 5.4% loss on Dogs to a 4.48% gain on Indoor, so the worst-case loss is larger than the headline number.

Reading between the lines

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

  • The method's success likely depends on how well OpenCLIP's frozen features separate fine-grained classes; on datasets where those features are weak, the binarized Bloom proxy would select irrelevant images and the accuracy trade-off would grow beyond 0.83%.
  • A testable extension is to replace the global zero-threshold binarization with learned per-dimension thresholds or multi-bit quantization, which could reduce false positives while keeping the fast hashing.
  • The same pattern—Bloom membership test followed by a small exact similarity search—could accelerate other large-scale subset-selection problems wherever a good pretrained feature encoder exists.
  • Because the paper does not analyze which downstream datasets lose the most accuracy, an important next step is to characterize the failure cases, e.g., whether losses concentrate on datasets with many visually similar classes.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 3 minor

Summary. The paper addresses coreset sampling for fine-grained self-supervised learning from an unlabeled open set. The authors propose BloomCoreset, which uses a pretrained OpenCLIP image encoder to extract 512-d features for the downstream dataset and the open set, binarizes those features by sign, inserts the downstream binary patterns into a Counting Bloom Filter, and then checks open-set patterns for membership. A top-k cosine-similarity filter over the real-valued features selects the final 1% coreset, which is used together with the downstream data to train SimCLR. The paper reports a 98.5% reduction in sampling time (from about 23 hours to about 20 minutes on two RTX 3090 GPUs) with a 0.83% average accuracy trade-off across 11 downstream datasets relative to SimCore.

Significance. The potential significance is high: coreset sampling from large unlabeled pools is a practical bottleneck, and a near-order-of-magnitude speedup with roughly 1% accuracy cost would make open-set SSL substantially more scalable. The paper evaluates across 11 downstream datasets and 3 open-sets, compares against the SimCore baseline, and includes a random-Bloom control in Table III, which is a useful breadth of experiments. However, the significance presently rests on the claim that the Bloom filter performs semantic similarity hashing. If, as Algorithm 1 indicates, the membership stage is a fixed-rate random prefilter, the contribution reduces to random subsampling plus cosine top-k, and the novelty claim is substantially weakened. The empirical results may still be valid for that variant, but the manuscript does not currently provide the evidence needed to separate these cases. The abstract's claim that the code is publicly available is also not backed by any repository link in the manuscript.

major comments (4)
  1. [II-B, Algorithm 1] In Algorithm 1, `cbloom.update(binz)` inserts the exact 512-bit sign pattern of each downstream image, and `cbloom.check(binz)` tests whether the same exact pattern was inserted. For any open-set image that is not a near-duplicate of a downstream image, the chance of exact match is at most NX/2^512, i.e., negligible. With k=10 hash functions and m=10000*NX/3500 bits, the false-positive rate is (1 - exp(-k*NX/m))^k = (1 - exp(-3.5))^10 ≈ 0.74, independent of image content. Therefore the ~74% pass rate implied by the setup consists almost entirely of false positives, and the Bloom stage is a content-independent Bernoulli(0.74) prefilter. The actual selection is done by the subsequent cosine top-k filter over the randomly retained 74%. This contradicts the paper's narrative in Section II and Fig. 2 that Bloom filters store both low- and high-level features and perform similarity-based membership. The authors must either specify a different update/check semantics that makes the membership test similarity-aware (e.g., hashing overlapping feature blocks or using multi-probe LSH), or explicitly state that the contribution is randomized prefiltering plus top-k selection and remove the Bloom-based semantic-retrieval claims. This issue is load-bearing for the central claim of the paper.
  2. [Table II] Table II reports the Cars accuracy trade-off as -5.1%, but the table values are 52.08 (ours) minus 57.90 (SimCore), which is -5.82%. The abstract's average trade-off of -0.83% is consistent with the -5.82% value and not with the tabulated -5.1%. This numerical inconsistency must be corrected; it directly affects the headline quantitative claim.
  3. [III-A and III-B, Tables II-III] No error bars or multiple seeds are reported for any accuracy number. The differences that support the trade-off claim are often small (e.g., Pet -0.7%, Food +0.42%, Faces -3.13%), and single-run fine-grained accuracy can easily shift by more than one percentage point across training seeds. The authors should report mean and standard deviation over at least 3 seeds for the 1% coreset experiments for both SimCore and BloomCoreset, or otherwise show that the reported trade-off is not within training noise.
  4. [Section II-B and abstract] The implementation is under-specified and the promised code is absent. The abstract states 'We have made the code publicly available,' but no repository link is provided in the manuscript. Algorithm 1 does not define the `CBloomFilter` operations (how murmurhash3 is used to derive the 10 hash functions, what `csize=32` controls, how the counting mechanism is used), nor the exact top-k rule of the `Filter` function beyond the budget B. The timing protocol for Table I is also incomplete: it does not state which OpenCLIP model was used, how many open-set images were processed, whether OpenCLIP feature-extraction time is included in the 20 minutes, or how the 23-hour SimCore number was obtained. Without these details the 98.5% speedup cannot be reproduced or audited.
minor comments (3)
  1. [Sections III and V] The method is called 'BloomSSL' in Section III and 'BloomCoreset' elsewhere, and the conclusion misspells it as 'BloomCorest'; the naming should be unified throughout.
  2. [Figure 3] The caption and legend of Figure 3 do not define the 'Ours' bar clearly; the caption should specify that 'Ours' is BloomCoreset applied to the corresponding open-set, and the axes should be labeled consistently.
  3. [Figure 4] The density maps in Figure 4 have no axis labels, color scale, or explanation of how 'Clip Downstream', 'Clip Coreset', 'Pre-trained Downstream', and 'Pre-trained Coreset' were generated; this makes the figure difficult to interpret.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: reported accuracies are measured on held-out tasks, and no parameter or claim reduces by construction to the paper's inputs.

full rationale

BloomCoreset is an empirical sampling method, not a derivation of predictions from first principles. The claimed result is that coresets sampled with a Bloom-filter-based pipeline, followed by top-k cosine filtering, yield downstream classification accuracy close to SimCore while reducing sampling time. That accuracy is obtained by training SimCLR from scratch on the sampled coreset plus downstream data and then linearly evaluating on held-out labels; it is not computed from the same OpenCLIP features used for sampling, so the evaluation is not self-referential by construction. No parameter is fitted to the 11 downstream accuracies, and the paper includes an honest ablation (Table III) showing that raw Bloom-filter samples without top-k filtering degrade performance, which indicates the post-filtering step carries real empirical weight. The strongest concern in the manuscript is not circularity but an implementation-mechanism mismatch: Algorithm 1 inserts and checks exact 512-bit sign patterns, so true exact matches with open-set images are essentially impossible and most membership passes are Bloom-filter false positives; this would make the membership stage behave like a random subsampler rather than a semantic similarity filter. That is a correctness and reproducibility issue about whether the method works as described, not a case where the paper's output is equivalent to its input by definition. The paper also lacks a code link and a hyperparameter tuning analysis for filter size, counter bits, hash count, and binarization threshold; these omissions are reproducibility risks, not circularity. Because the central claim is externally falsifiable against held-out benchmarks and no load-bearing step reduces to a self-citation, a fit, or a renamed input, no circularity is found.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical or conceptual entities. It relies on existing pretrained models, an existing data structure, and several ad hoc hyperparameters. The main uncharged assumptions are about transferability of OpenCLIP features and the sufficiency of binarized hashing.

free parameters (5)
  • Bloom filter size hsize = 10000 * (NX/3500)
    Heuristically scaled with downstream set size in Algorithm 1; no sensitivity analysis or justification is provided.
  • Counter size csize = 32
    Chosen without explanation in Algorithm 1; affects storage and false positive rate.
  • Number of hash functions = 10
    The paper uses ten variations of murmurhash3; no analysis of the effect on false positives or accuracy is given.
  • Binarization threshold = 0
    OpenCLIP features are converted to 0/1 by thresholding at zero; this loses magnitude information and is not ablated.
  • Top-k budget B = 1% of Open-Set
    Inherited from SimCore as the experimental protocol, not derived from the method itself.
assumptions (4)
  • domain assumption OpenCLIP features are a reliable proxy for fine-grained semantic similarity.
    Section II-B and Fig. 1 use OpenCLIP to extract features for both downstream and open-set images; if these features miss fine-grained distinctions, the coreset will be poor.
  • domain assumption Binarizing features to signs preserves enough similarity structure for Bloom filter membership.
    Algorithm 1 thresholds features at zero; the paper does not quantify information lost by discarding magnitudes.
  • domain assumption Bloom filter false positives can be corrected by top-k cosine filtering.
    Section II-B asserts this; Table III shows it helps on 4 datasets, but the candidate set size and filter details are not analyzed.
  • domain assumption The SimCore reimplementation is faithful.
    Table I compares against a re-run of SimCore with batch size 256 on two GPUs; no code or hyperparameters are provided to verify fairness.

how reviews work

0 comments
Cite this review

Pith. "Pith review of BloomCoreset: Fast Coreset Sampling using Bloom Filters for Fine-Grained Self-Supervised Learning." pith.science (2026). https://pith.science/paper/SNYDUCR2

@misc{pith2026241216942,
  author       = {Pith},
  title        = {Pith review of: BloomCoreset: Fast Coreset Sampling using Bloom Filters for Fine-Grained Self-Supervised Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SNYDUCR2}},
  note         = {Machine review of arXiv:2412.16942}
}
abstract

The success of deep learning in supervised fine-grained recognition for domain-specific tasks relies heavily on expert annotations. The Open-Set for fine-grained Self-Supervised Learning (SSL) problem aims to enhance performance on downstream tasks by strategically sampling a subset of images (the Core-Set) from a large pool of unlabeled data (the Open-Set). In this paper, we propose a novel method, BloomCoreset, that significantly reduces sampling time from Open-Set while preserving the quality of samples in the coreset. To achieve this, we utilize Bloom filters as an innovative hashing mechanism to store both low- and high-level features of the fine-grained dataset, as captured by Open-CLIP, in a space-efficient manner that enables rapid retrieval of the coreset from the Open-Set. To show the effectiveness of the sampled coreset, we integrate the proposed method into the state-of-the-art fine-grained SSL framework, SimCore [1]. The proposed algorithm drastically outperforms the sampling strategy of the baseline in SimCore [1] with a $98.5\%$ reduction in sampling time with a mere $0.83\%$ average trade-off in accuracy calculated across $11$ downstream datasets.

Figures

Figures reproduced from arXiv: 2412.16942 by the authors.

Figure 2
Figure 2. BloomCoreset. Domain-specific features are used to build the Counting Bloom Filter (CBF). A membership test is then conducted to sample Open-Set images similar to domain￾specific ones. After sampling, the inner product is calculated, and additional filtering is applied to select the best subset from the sampled Open-Set images. A detailed overview of the sampling method is provided in Section II-B. A. Problem Formul… view at source ↗
Figure 3
Figure 3. Other Open-Sets. While comparing across different Open-Sets, the learned representation from the coreset sampled using the proposed method shows competitive performance to the baseline and, in some cases, outperforms the baseline. we compute cosine similarity scores between the filtered and downstream samples to retrieve Open-Set samples most similar to the downstream dataset (see Algorithm 1). The entire sampling p… view at source ↗
Figure 4
Figure 4. Feature Distribution. Show representation space of Open-CLIP [25]. We have used Gaussian kernel density estimation [43] to show the feature distribution of downstream dataset samples and coreset samples across the unit ring. IV. DISCUSSION Table I demonstrates that our proposed algorithm signifi￾cantly reduces sampling time by 98.5% compared to [1], with only a 0.83% average tradeoff in accuracy across 11 datasets. … view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 30 canonical work pages

  1. [1]

    Coreset sampling from open-set for fine-grained self-supervised learning,

    S. Kim, S. Bae, and S.-Y . Yun, “Coreset sampling from open-set for fine-grained self-supervised learning,” in IEEE/CVF CVPR , 2023, pp. 7537–7547

    work page 2023

  2. [2]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in IEEE CVPR, 2016, pp. 770–778

    2016

  3. [3]

    Segnet: A deep convolutional encoder-decoder architecture for image segmentation,

    V . Badrinarayanan, A. Kendall, and R. Cipolla, “Segnet: A deep convolutional encoder-decoder architecture for image segmentation,” IEEE TPAMI, vol. 39, no. 12, pp. 2481–2495, 2017

    work page 2017

  4. [4]

    Fully convolutional networks for semantic segmentation,

    J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in IEEE CVPR, 2015, pp. 3431–3440

    work page 2015

  5. [5]

    Photographic image synthesis with cascaded refinement networks,

    Q. Chen and V . Koltun, “Photographic image synthesis with cascaded refinement networks,” in IEEE ICCV, 2017, pp. 1511–1520

    work page 2017

  6. [6]

    High- resolution image synthesis with latent diffusion models,

    R. Rombach, A. Blattmann, D. Lorenz, P. Esser, and B. Ommer, “High- resolution image synthesis with latent diffusion models,” in IEEE/CVF CVPR, 2022, pp. 10 684–10 695

    work page 2022

  7. [7]

    Pointnet: Deep learning on point sets for 3d classification and segmentation,

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas, “Pointnet: Deep learning on point sets for 3d classification and segmentation,” in IEEE CVPR, 2017, pp. 652–660

    work page 2017

  8. [8]

    Stickypillars: Robust and efficient feature matching on point clouds using graph neural networks,

    K. Fischer, M. Simon, F. Olsner, S. Milz, H.-M. Gross, and P. Mader, “Stickypillars: Robust and efficient feature matching on point clouds using graph neural networks,” in IEEE/CVF CVPR, 2021, pp. 313–323

    work page 2021

Show all 43 references
  1. [9]

    Extreme consistency: Overcoming annotation scarcity and domain shifts,

    G. Fotedar, N. Tajbakhsh, S. Ananth, and X. Ding, “Extreme consistency: Overcoming annotation scarcity and domain shifts,” in Medical Image Computing and Computer Assisted Intervention–MICCAI 2020: 23rd International Conference, Lima, Peru, October 4–8, 2020, Proceedings, Part...

    2020

  2. [10]

    Self-supervised learning methods and applications in medical imaging analysis: A survey,

    S. Shurrab and R. Duwairi, “Self-supervised learning methods and applications in medical imaging analysis: A survey,” PeerJ Computer Science, vol. 8, p. e1045, 2022

    2022

  3. [11]

    Advances in deep learning models for resolving medical image segmentation data scarcity problem: A topical review,

    A. K. Upadhyay and A. K. Bhandari, “Advances in deep learning models for resolving medical image segmentation data scarcity problem: A topical review,” Archives of Computational Methods in Engineering , vol. 31, no. 3, pp. 1701–1719, 2024

    2024

  4. [12]

    Imagenet: A large-scale hierarchical image database,

    J. Deng, W. Dong, R. Socher, L.-J. Li, K. Li, and L. Fei-Fei, “Imagenet: A large-scale hierarchical image database,” in 2009 IEEE CVPR . Ieee, 2009, pp. 248–255

    2009

  5. [13]

    Bootstrap your own latent-a new approach to self-supervised learning,

    J.-B. Grill, F. Strub, F. Altch ´e, C. Tallec, P. Richemond, E. Buchatskaya, C. Doersch, B. Avila Pires, Z. Guo, M. Gheshlaghi Azar et al., “Bootstrap your own latent-a new approach to self-supervised learning,” Advances in neural information processing systems , vol. 33, pp. ...

    2020

  6. [14]

    Emerging properties in self-supervised vision transformers,

    M. Caron, H. Touvron, I. Misra, H. J ´egou, J. Mairal, P. Bojanowski, and A. Joulin, “Emerging properties in self-supervised vision transformers,” in IEEE/CVF ICCV, 2021, pp. 9650–9660

    2021

  7. [15]

    Unsupervised representation learning by predicting image rotations,

    S. Gidaris, P. Singh, and N. Komodakis, “Unsupervised representation learning by predicting image rotations,” arXiv preprint arXiv:1803.07728, 2018

    2018 arXiv

  8. [16]

    Colorful image colorization,

    R. Zhang, P. Isola, and A. A. Efros, “Colorful image colorization,” in Computer Vision–ECCV 2016: 14th European Conference, Amsterdam, The Netherlands, October 11-14, 2016, Proceedings, Part III 14 . Springer, 2016, pp. 649–666

    2016

  9. [17]

    A simple framework for contrastive learning of visual representations,

    T. Chen, S. Kornblith, M. Norouzi, and G. Hinton, “A simple framework for contrastive learning of visual representations,” in ICML. PMLR, 2020, pp. 1597–1607

    2020

  10. [18]

    Representation learning with contrastive predictive coding,

    A. v. d. Oord, Y . Li, and O. Vinyals, “Representation learning with contrastive predictive coding,” arXiv preprint arXiv:1807.03748 , 2018

    2018 arXiv

  11. [19]

    Self-supervised pretraining of visual features in the wild,

    P. Goyal, M. Caron, B. Lefaudeux, M. Xu, P. Wang, V . Pai, M. Singh, V . Liptchinsky, I. Misra, A. Joulin et al., “Self-supervised pretraining of visual features in the wild,” arXiv preprint arXiv:2103.01988 , 2021

    2021 arXiv

  12. [20]

    How well do self- supervised models transfer?

    L. Ericsson, H. Gouk, and T. M. Hospedales, “How well do self- supervised models transfer?” in IEEE/CVF CVPR, 2021, pp. 5414–5423

    2021

  13. [21]

    Active learning for convolutional neural networks: A core-set approach,

    O. Sener and S. Savarese, “Active learning for convolutional neural networks: A core-set approach,” arXiv preprint arXiv:1708.00489 , 2017

    2017 arXiv

  14. [22]

    Submodularity in data subset selection and active learning,

    K. Wei, R. Iyer, and J. Bilmes, “Submodularity in data subset selection and active learning,” in ICML. PMLR, 2015, pp. 1954–1963

    2015

  15. [23]

    Summary cache: a scalable wide-area web cache sharing protocol,

    L. Fan, P. Cao, J. Almeida, and A. Z. Broder, “Summary cache: a scalable wide-area web cache sharing protocol,” IEEE/ACM transactions on networking, vol. 8, no. 3, pp. 281–293, 2000

    2000

  16. [24]

    Space/time trade-offs in hash coding with allowable errors,

    B. H. Bloom, “Space/time trade-offs in hash coding with allowable errors,” Communications of the ACM , vol. 13, no. 7, pp. 422–426, 1970

    1970

  17. [25]

    Reproducible scaling laws for contrastive language-image learning,

    M. Cherti, R. Beaumont, R. Wightman, M. Wortsman, G. Ilharco, C. Gordon, C. Schuhmann, L. Schmidt, and J. Jitsev, “Reproducible scaling laws for contrastive language-image learning,” in IEEE/CVF CVPR, 2023, pp. 2818–2829

    2023

  18. [26]

    LAION-5b: An open large- scale dataset for training next generation image-text models,

    C. Schuhmann, R. Beaumont, R. Vencu, C. W. Gordon, R. Wightman, M. Cherti, T. Coombes, A. Katta, C. Mullis, M. Wortsman, P. Schramowski, S. R. Kundurthy, K. Crowson, L. Schmidt, R. Kaczmarczyk, and J. Jitsev, “LAION-5b: An open large- scale dataset for training next generation...

    2022

  19. [27]

    Learning transferable visual models from natural language supervision,

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, G. Krueger, and I. Sutskever, “Learning transferable visual models from natural language supervision,” in ICML, 2021

    2021

  20. [28]

    Murmurhash3,

    A. Appleby, “Murmurhash3,” 2016, https://github.com/aappleby/ smhasher/wiki/MurmurHash3 [Accessed: (January 14, 2024)]

    2016

  21. [29]

    Fine- grained visual classification of aircraft,

    S. Maji, E. Rahtu, J. Kannala, M. Blaschko, and A. Vedaldi, “Fine- grained visual classification of aircraft,” arXiv preprint arXiv:1306.5151 , 2013

    2013 arXiv

  22. [30]

    3d object representations for fine-grained categorization,

    J. Krause, M. Stark, J. Deng, and L. Fei-Fei, “3d object representations for fine-grained categorization,” in IEEE ICCV workshops , 2013, pp. 554–561

    2013

  23. [31]

    Cats and dogs,

    O. M. Parkhi, A. Vedaldi, A. Zisserman, and C. Jawahar, “Cats and dogs,” in 2012 IEEE conference on computer vision and pattern recognition . IEEE, 2012, pp. 3498–3505

    2012

  24. [32]

    Caltech-UCSD Birds 200,

    P. Welinder, S. Branson, T. Mita, C. Wah, F. Schroff, S. Belongie, and P. Perona, “Caltech-UCSD Birds 200,” California Institute of Technology, Tech. Rep. CNS-TR-2010-001, 2010

    2010

  25. [33]

    Novel dataset for fine-grained image categorization: Stanford dogs,

    A. Khosla, N. Jayadevaprakash, B. Yao, and F.-F. Li, “Novel dataset for fine-grained image categorization: Stanford dogs,” in Proc. CVPR workshop on fine-grained visual categorization (FGVC), vol. 2. Citeseer, 2011

    2011

  26. [34]

    Automated flower classification over a large number of classes,

    M.-E. Nilsback and A. Zisserman, “Automated flower classification over a large number of classes,” in 2008 Sixth Indian conference on computer vision, graphics & image processing . IEEE, 2008, pp. 722–729

    2008

  27. [35]

    Human action recognition by learning bases of action attributes and parts,

    B. Yao, X. Jiang, A. Khosla, A. L. Lin, L. Guibas, and L. Fei-Fei, “Human action recognition by learning bases of action attributes and parts,” in 2011 International conference on computer vision . IEEE, 2011, pp. 1331–1338

    2011

  28. [36]

    Recognizing indoor scenes,

    A. Quattoni and A. Torralba, “Recognizing indoor scenes,” in 2009 IEEE conference on computer vision and pattern recognition . IEEE, 2009, pp. 413–420

    2009

  29. [37]

    Describing textures in the wild,

    M. Cimpoi, S. Maji, I. Kokkinos, S. Mohamed, and A. Vedaldi, “Describing textures in the wild,” in IEEE CVPR, 2014, pp. 3606–3613

    2014

  30. [38]

    Maskgan: Towards diverse and interactive facial image manipulation,

    C.-H. Lee, Z. Liu, L. Wu, and P. Luo, “Maskgan: Towards diverse and interactive facial image manipulation,” in IEEE/CVF CVPR, 2020, pp. 5549–5558

    2020

  31. [39]

    Food/non-food image classifi- cation and food categorization using pre-trained googlenet model,

    A. Singla, L. Yuan, and T. Ebrahimi, “Food/non-food image classifi- cation and food categorization using pre-trained googlenet model,” in Proceedings of the 2nd International Workshop on Multimedia Assisted Dietary Management, 2016, pp. 3–11

    2016

  32. [40]

    Microsoft coco: Common objects in context,

    T.-Y . Lin, M. Maire, S. Belongie, J. Hays, P. Perona, D. Ramanan, P. Doll´ar, and C. L. Zitnick, “Microsoft coco: Common objects in context,” in Computer Vision–ECCV 2014: 13th European Conference, Zurich, Switzerland, September 6-12, 2014, Proceedings, Part V 13 . Springer, ...

    2014

  33. [41]

    The inaturalist species classifica- tion and detection dataset,

    G. Van Horn, O. Mac Aodha, Y . Song, Y . Cui, C. Sun, A. Shepard, H. Adam, P. Perona, and S. Belongie, “The inaturalist species classifica- tion and detection dataset,” in IEEE CVPR, 2018, pp. 8769–8778

    2018

  34. [42]

    False negative problem of counting bloom filter,

    D. Guo, Y . Liu, X. Li, and P. Yang, “False negative problem of counting bloom filter,” IEEE Transactions on Knowledge and Data Engineering , vol. 22, no. 5, pp. 651–664, 2010

    2010

  35. [43]

    Understanding contrastive representation learning through alignment and uniformity on the hypersphere,

    T. Wang and P. Isola, “Understanding contrastive representation learning through alignment and uniformity on the hypersphere,” in ICML. PMLR, 2020, pp. 9929–9939

    2020

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.