Pith. sign in

REVIEW 4 major objections 4 minor 62 references

Assessing Pre-Trained Models for Transfer Learning Through Distribution of Spectral Components

T0 review · 4 major / 4 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 Pre-trained models can be ranked for transfer learning by the distribution of their features' singular values, weighted by each spectral component's fit to target labels.

desk verdict DISCO adds a genuinely new SVD-based transferability metric with solid empirical wins, but the core weighting premise is verified only on CNNs and the SOTA claims rest on thin margins. read the letter →

arxiv 2412.19085 v2 pith:H7WWLXUP submitted 2024-12-26 cs.LG

Tengxue Zhang , Yang Shu , Xinyang Chen , Yifei Long , Chenjuan Guo , Bin Yang This is my paper · ORCID
classification cs.LG
keywords transferlearningpre-trainedmodelselectiontransferabilityestimationsingularvaluedecompositionspectralcomponentshubfine-tuningrankingcorrelation
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

Pre-trained models in a model hub differ wildly in how well they transfer to a new task, and fine-tuning every candidate to find out is expensive. This paper claims that the fine-tuning process itself leaves a fingerprint in the singular-value spectrum of the extracted features: after fine-tuning, the singular values concentrate on a few spectral components, and those components are the transferable ones. The authors build a score, DISCO, that weights each spectral component by its ratio of singular values and its estimated performance on the target labels, and they report that this score ranks pre-trained models better than existing transferability metrics on supervised classification (average weighted Kendall's tau 0.739), self-supervised classification (0.726), and object detection (0.634). If the claim holds, model selection becomes a fast forward pass plus a cheap linear or centroid estimate instead of repeated fine-tuning.

What carries the argument

The central object is the SVD of the feature matrix $Z = U\Sigma V^T$ of target data extracted by each pre-trained model, with singular values grouped into $G$ spectral components. The load-bearing identity is the score $S_{\mathrm{DISCO}} = \sum_g S_g^{\mathrm{task}} S_g^{\mathrm{ratio}}$, where $S_g^{\mathrm{ratio}}$ is the proportion of total singular values in component $g$ and $S_g^{\mathrm{task}}$ is a task-specific transferability estimate (nearest-centroid confidence for classification, negative MSE of a linear-regression pseudo-inverse for regression). Hard-example selection via LDA reduces the sample size so the SVD stays cheap. This machinery does the work of turning the empirical concentration observation into a ranking criterion.

What would settle it

Compute $S_g^{\mathrm{ratio}}$ before and after fine-tuning for a model pool that includes vision transformers or large domain shifts; if the singular-value proportion does not consistently increase for the largest components, or if the DISCO ranking's weighted Kendall's tau against true fine-tuning performance drops to chance levels, the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that a pre-trained model's transferability to a downstream task is determined by the distribution of spectral components of its extracted features, not by properties of the whole feature set. The paper observes, through SVD of features before and after fine-tuning, that fine-tuning concentrates the singular-value distribution onto components with larger singular values, and that these components transfer better. DISCO formalizes this as $S_{\mathrm{DISCO}} = \sum_{g=1}^{G} S_g^{\mathrm{task}} \cdot S_g^{\mathrm{ratio}}$, with $S_g^{\mathrm{task}}$ measuring how well the $g$-th component separates target classes (nearest-centroid confidence) or fits target regression labels (linear least squares), and $S_g^{\mathrm{ratio}}$ the share of total singular values in that component. The paper reports that this score achieves the highest average weighted Kendall's tau among compared metrics on all three benchmarks.

Load-bearing premise

The paper assumes that the observed tendency of fine-tuning to concentrate singular values onto the largest spectral components, measured on three CNN architectures and three image datasets, is a general property of fine-tuning across architectures, tasks, and domain gaps.

Editorial extensions

If this is right

  • Model hubs can be ranked with a single forward pass plus a lightweight classifier, avoiding per-model fine-tuning.
  • The same framework works for classification and regression, so object detection and other multi-task settings can share one selection criterion.
  • Self-supervised models, which lack pre-trained classifiers, can still be ranked because the score uses only features and target labels.
  • With hard-example selection, rankings remain strong at 20 to 40 percent of the data, making the method practical for large target datasets.
  • The score is a weighted sum of interpretable per-component terms, so a model's transferability profile can be inspected rather than reduced to a single black-box number.

Reading between the lines

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

  • If the spectral-concentration pattern holds for transformer backbones, DISCO could be applied to large model hubs of foundation models essentially unchanged, since it only needs features and labels.
  • The per-component weighting suggests a testable extension: instead of choosing one model, one could blend spectral components across models, selecting the best component per group to form an ensemble.
  • The hard-example sampling insight, that hard examples carry more information about transferability, could be reused by other transferability metrics, since the paper shows random sampling loses more accuracy at equal sample sizes.
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

4 major / 4 minor

Summary. The paper proposes DISCO, a transferability metric for pre-trained model selection. It performs SVD on features extracted by each candidate model, groups the singular values into spectral components, and scores each component with a task-specific measure (a nearest-centroid classifier score for classification, a linear-regression score for object detection) weighted by the component's singular-value share (Eq. 6, Eq. 10, Eq. 13). The authors motivate the weighting by an empirical observation that fine-tuning concentrates singular values onto larger spectral components (Fig. 2, App. B.1). DISCO is evaluated with weighted Kendall's tau against ground-truth fine-tuning results on 11 classification datasets for supervised and self-supervised CNN pools and on five object detection datasets with six YOLO variants, reporting average tau values of 0.739, 0.726, and 0.634, respectively, and is presented as achieving state-of-the-art model-selection performance.

Significance. If the empirical premise holds, DISCO is a simple, label-based, fine-tuning-free scoring method that extends naturally from classification to regression-style tasks, and the hard-example selection in Sec. 3.4 addresses practical scalability. The paper's strengths are its clean SVD decomposition, the task-specific score design, and the breadth of classification benchmarks, including both supervised and self-supervised model pools. The appendix's extended benchmark and tie-correction analysis are honest additions, but they also expose the main unresolved issue: the load-bearing spectral-concentration premise is verified only on a small set of CNNs and is not checked on transformers where the method's average correlation drops substantially.

major comments (4)
  1. [3.2, Eq. (6); Fig. 2; App. B.1 and B.5, Table 9] The rating weight Sg_ratio in Eq. (6), which is the only new ingredient distinguishing DISCO from a plain sum of per-component task scores, is justified entirely by the empirical claim that fine-tuning concentrates singular values onto larger spectral components. That claim is demonstrated in Fig. 2 and App. B.1 on exactly three CNN backbones (ResNet-152, DenseNet-201, Inception-v3) and three datasets (Caltech101, Flowers, Pets). The extended benchmark in App. B.5 includes ViT-S, ViT-B, and Swin-T, but it reports only ranking correlations: the average tau drops from 0.739 in Table 1 to 0.554 in Table 9. Because the spectral-concentration diagnostic for transformers is never shown, the premise of Eq. (6) remains unverified outside the CNN family, and a failure of that premise could misrank models even when the per-component task scores are accurate. Please report the CF and Sratio plots for the three transformer models in App. B.5 and provide an ablation on the extended pool that replaces Sg_ratio with uniform weights or with architecture-specific weights.
  2. [4.1, Tables 1 and 2] The claimed state-of-the-art result is not supported by any uncertainty quantification or significance test. The average advantage over NCTI is only 0.013 on supervised models (0.739 vs. 0.726) and 0.007 on self-supervised models (0.726 vs. 0.719), and the per-dataset signs are mixed: for example, in Table 1 DISCO scores 0.575 on Pets where NCTI scores 0.924, and in Table 2 DISCO scores 0.542 on Pets where NCTI scores 0.805. Since each table entry appears to be a single deterministic computation, a paired bootstrap or permutation test across the 11 datasets would clarify whether the average gain is reliable. Without this, the headline 'state-of-the-art' claim is fragile.
  3. [4.3, Table 3] The object-detection experiment is too narrow to support the paper's generality claim. The pool consists of six YOLOv5/YOLOv8 variants from the same detector family, all pre-trained on COCO, and the five datasets yield only 15 pairwise comparisons per dataset, making tau values and top-k probabilities coarse. The perfect Pr(top1)=1.000 on this small pool is not strong evidence that the regression score in Eq. (12) generalizes to diverse detection architectures. Please extend the pool to at least one non-YOLO family (e.g., Faster R-CNN or a transformer detector) or temper the claim of generality accordingly.
  4. [3.4] The complexity statement for naive SVD is incorrect. For a matrix Z in R^{N x d}, standard SVD complexity is O(N d^2) when N > d, not O(N^3 d) as stated. After PCA reduction to d=128 in App. A.2, the relevant tall-skinny regime indeed gives O(N d^2). The empirical runtime results in Fig. 4d and Table 5 may still be valid, but the analytical motivation for hard-example selection should be corrected and re-derived.
minor comments (4)
  1. [3.2 and 4.4, Fig. 4c] The main experiments use G=10 (Sec. 3.2), but Sec. 4.4 states that G=6 or G=8 'offer the best balance.' Please clarify whether Tables 1-3 use G=10 throughout and whether the reported averages change if the recommended G is used; this affects how the reader interprets the group-number sensitivity.
  2. [App. B.6, Table 10] The tie-correction analysis considers models within 0.1% accuracy of each other as tied, but it reports only DISCO vs. SFDA and LogME, omitting NCTI and ETran, which are the relevant competitors in Tables 1 and 3. Please include those baselines or explain the omission.
  3. [Throughout] Several small typographical errors should be fixed, including 'framawork' in the caption of Fig. 4a and 'featrue' in the text surrounding Eq. (16).
  4. [4.3 and App. B.4, Table 8] The object-detection ground-truth table does not state the training protocol (epochs, batch size, augmentation, hyperparameter sweep) used to obtain the mAP50 values. Adding these details would make the benchmark reproducible.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: DISCO's score is constructed from pre-fine-tuning SVD components and downstream labels, with fine-tuning performance used only for evaluation.

full rationale

The derivation chain is self-contained. DISCO's score (Eq. 6) is a weighted sum S_DISCO = sum_g S_g_task * S_g_ratio, where S_g_ratio (Eq. 4) is computed from singular values of the original pre-fine-tuning feature matrix Z, and S_g_task is either a nearest-centroid classifier accuracy on the g-th spectral component (Eqs. 7-9) or a negative MSE of a pseudo-inverse linear fit (Eqs. 11-12). No parameter is fitted to the ground-truth fine-tuning accuracies {P^m}; those accuracies enter only in the evaluation protocol (weighted Kendall's tau, Appendix A.1) after the metric has been computed. The empirical observation that fine-tuning concentrates singular values onto larger components (Fig. 2, Appendix B.1) is used as motivation for the S_g_ratio weighting, not as a fitted calibration: the weight is not chosen to optimize tau on the benchmarks. Neither is any load-bearing result imported from the authors' own prior work; the cited sources for fine-tuning ground truth (Shao et al. 2022) and benchmark protocols are external and used for data or evaluation, not to justify the construction. The extended-benchmark and robustness analyses (Tables 9-10) are additional checks, not part of the score definition. The concern that the spectral-concentration premise is validated on only three CNNs and three datasets is a generality or robustness limitation, not a circularity: it does not make Eq. 6 equivalent to its inputs by construction. Therefore no circular step is present.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim (DISCO ranks models well) rests on an empirical generalization from a small fine-tuning study, on Gaussian and linearity assumptions for the per-component scores, and on hand-chosen hyperparameters (G and PCA dimension). No invented entities are introduced; spectral components are a rearrangement of existing feature dimensions.

free parameters (2)
  • G (number of spectral groups) = 10 in main experiments; 6 or 8 reported as better in Figure 4c
    The number of groups determines how finely the spectral components are split. It is chosen by hand and affects the final score; the paper uses G=10 for reported results while recommending 6 or 8.
  • PCA dimension for features = 128
    Features are projected to 128 dimensions before SVD (Appendix A.2). This choice affects all scores and is not derived from data.
assumptions (3)
  • domain assumption Empirical generalization: after fine-tuning, singular values concentrate on components with larger singular values, and these components are more transferable.
    Observed on 3 models and 3 datasets (Figure 2, Appendix B.1), assumed to hold across the model pools and tasks evaluated; motivates the Sg_ratio weighting in Eq. 6.
  • domain assumption Multivariate Gaussian class-conditional distributions for features within each spectral component.
    Used in Eq. 7 to define the nearest centroid classifier for the classification score Sg_ncc; no normality test is provided.
  • domain assumption Linear or least-squares fit of features to labels is a sufficient proxy for fine-tuned regression performance.
    Used to define Sg_lr in Eq. 12 for object detection; assumes that a linear readout on each component reflects its fine-tuning value.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Assessing Pre-Trained Models for Transfer Learning Through Distribution of Spectral Components." pith.science (2026). https://pith.science/paper/H7WWLXUP

@misc{pith2026241219085,
  author       = {Pith},
  title        = {Pith review of: Assessing Pre-Trained Models for Transfer Learning Through Distribution of Spectral Components},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H7WWLXUP}},
  note         = {Machine review of arXiv:2412.19085}
}
read the original abstract

Pre-trained model assessment for transfer learning aims to identify the optimal candidate for the downstream tasks from a model hub, without the need of time-consuming fine-tuning. Existing advanced works mainly focus on analyzing the intrinsic characteristics of the entire features extracted by each pre-trained model or how well such features fit the target labels. This paper proposes a novel perspective for pre-trained model assessment through the Distribution of Spectral Components (DISCO). Through singular value decomposition of features extracted from pre-trained models, we investigate different spectral components and observe that they possess distinct transferability, contributing diversely to the fine-tuning performance. Inspired by this, we propose an assessment method based on the distribution of spectral components which measures the proportions of their corresponding singular values. Pre-trained models with features concentrating on more transferable components are regarded as better choices for transfer learning. We further leverage the labels of downstream data to better estimate the transferability of each spectral component and derive the final assessment criterion. Our proposed method is flexible and can be applied to both classification and regression tasks. We conducted comprehensive experiments across three benchmarks and two tasks including image classification and object detection, demonstrating that our method achieves state-of-the-art performance in choosing proper pre-trained models from the model hub for transfer learning.

Figures

Figures reproduced from arXiv: 2412.19085 by the authors.

Figure 1
Figure 1. The overall framework for assessing pre-trained [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The relative changes of Frobenius norm CF and the proportion of singular values Sratio in different spectral compo￾nents of extracted features before and after fine-tuning. {S m}M m=1 for the model hub should strongly correlate with their actual fine-tuning performance {P m}M m=1, allowing us to identify the best model for transfer. 3.2 Transferability Assessment Framework through Spectral Component Distribution To … view at source ↗
Figure 3
Figure 3. Overview of DISCO’s framework (better viewed in color). [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: (a) Ablation study on the framework and (b) on the object detection benchmark. (c) The average [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The relative changes of Frobenius norm CF and the proportion of singular values Sratio in different spectral compo￾nents of extracted features before and after fine-tuning. dataset Aircraft Caltech Cars Cifar10 Cifar100 DTD Flowers Food Pets SUN VOC Average 20% random …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

62 extracted references · 42 canonical work pages

  1. [1]

    S.; Davis, A.; Dean, J.; Devin, M.; et al

    Abadi, M.; Agarwal, A.; Barham, P.; Brevdo, E.; Chen, Z.; Citro, C.; Corrado, G. S.; Davis, A.; Dean, J.; Devin, M.; et al. 2016. Tensorflow: Large-scale machine learning on heterogeneous distributed systems. arXiv preprint arXiv:1603.04467

    arXiv 2016

  2. [2]

    M.; Rupprecht, C.; and Vedaldi, A

    Asano, Y. M.; Rupprecht, C.; and Vedaldi, A. 2019. Self-labelling via simultaneous clustering and representation learning. arXiv preprint arXiv:1911.05371

    arXiv 2019

  3. [3]

    ASD. 2022. wlots Dataset. \ https://universe.roboflow.com/asd-culfr/wlots . Accessed: 2023-01-27

    work page 2022

  4. [4]

    Bolya, D.; Mittapalli, R.; and Hoffman, J. 2021. Scalable diverse model selection for accessible transfer learning. In ‌NeurIPS

    work page 2021

  5. [5]

    Bossard, L.; Guillaumin, M.; and Van Gool, L. 2014. Food-101--mining discriminative components with random forests. In ECCV

    work page 2014

  6. [6]

    D.; Dhariwal, P.; Neelakantan, A.; Shyam, P.; Sastry, G.; Askell, A.; et al

    Brown, T.; Mann, B.; Ryder, N.; Subbiah, M.; Kaplan, J. D.; Dhariwal, P.; Neelakantan, A.; Shyam, P.; Sastry, G.; Askell, A.; et al. 2020. Language models are few-shot learners. In ‌NeurIPS

    work page 2020

  7. [7]

    Caron, M.; Bojanowski, P.; Joulin, A.; and Douze, M. 2018. Deep clustering for unsupervised learning of visual features. In ECCV

    work page 2018

  8. [8]

    Caron, M.; Misra, I.; Mairal, J.; Goyal, P.; Bojanowski, P.; and Joulin, A. 2020. Unsupervised learning of visual features by contrasting cluster assignments. In ‌NeurIPS

    work page 2020

Show all 62 references
  1. [9]

    Chen, X.; Fan, H.; Girshick, R.; and He, K. 2020. Improved baselines with momentum contrastive learning. arXiv preprint arXiv:2003.04297

    2020 arXiv

  2. [10]

    Cimpoi, M.; Maji, S.; Kokkinos, I.; Mohamed, S.; and Vedaldi, A. 2014. Describing textures in the wild. In CVPR

    2014

  3. [11]

    Dettmers, T.; Pagnoni, A.; Holtzman, A.; and Zettlemoyer, L. 2024. Qlora: Efficient finetuning of quantized llms. In ‌NeurIPS

    2024

  4. [12]

    Devlin, J.; Chang, M.-W.; Lee, K.; and Toutanova, K. 2018. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805

    2018 arXiv

  5. [13]

    Ding, N.; Chen, X.; Levinboim, T.; Changpinyo, S.; and Soricut, R. 2022. Pactran: Pac-bayesian metrics for estimating the transferability of pretrained models to classification tasks. In ECCV

    2022

  6. [14]

    Ding, Y.; Jiang, B.; Yu, A.; Zheng, A.; and Liang, J. 2024. Which Model to Transfer? A Survey on Transferability Estimation. arXiv preprint arXiv:2402.15231

    2024 arXiv

  7. [15]

    Donahue, J.; Jia, Y.; Vinyals, O.; Hoffman, J.; Zhang, N.; Tzeng, E.; and Darrell, T. 2014. Decaf: A deep convolutional activation feature for generic visual recognition. In ICML

    2014

  8. [16]

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

    2020 arXiv

  9. [17]

    Erhan, D.; Courville, A.; Bengio, Y.; and Vincent, P. 2010. Why does unsupervised pre-training help deep learning? In AISTATS

    2010

  10. [18]

    K.; Winn, J.; and Zisserman, A

    Everingham, M.; Van Gool, L.; Williams, C. K.; Winn, J.; and Zisserman, A. 2010. The pascal visual object classes (voc) challenge. IJCV, 88: 303--338

    2010

  11. [19]

    Fei-Fei, L.; Fergus, R.; and Perona, P. 2004. Learning generative visual models from few training examples: An incremental bayesian approach tested on 101 object categories. In CVPR

    2004

  12. [20]

    Gholami, M.; Akbari, M.; Wang, X.; Kamranian, B.; and Zhang, Y. 2023. Etran: Energy-based transferability estimation. In ICCV

    2023

  13. [21]

    Grill, J.-B.; Strub, F.; Altch \'e , F.; Tallec, C.; Richemond, P.; Buchatskaya, E.; Doersch, C.; Avila Pires, B.; Guo, Z.; Gheshlaghi Azar, M.; et al. 2020. Bootstrap your own latent-a new approach to self-supervised learning. In ‌NeurIPS

    2020

  14. [22]

    He, K.; Fan, H.; Wu, Y.; Xie, S.; and Girshick, R. 2020. Momentum contrast for unsupervised visual representation learning. In CVPR

    2020

  15. [23]

    He, K.; Zhang, X.; Ren, S.; and Sun, J. 2016. Deep residual learning for image recognition. In CVPR

    2016

  16. [24]

    home. 2022. NFL-competition Dataset. \ https://universe.roboflow.com/home-mxzv1/nfl-competition . Accessed: 2023-01-18

    2022

  17. [25]

    Huang, G.; Liu, Z.; Van Der Maaten, L.; and Weinberger, K. Q. 2017. Densely connected convolutional networks. In CVPR

    2017

  18. [26]

    Jia, M.; Tang, L.; Chen, B.-C.; Cardie, C.; Belongie, S.; Hariharan, B.; and Lim, S.-N. 2022. Visual prompt tuning. In ECCV

    2022

  19. [27]

    Jocher, G.; Chaurasia, A.; and Qiu, J. 2023. YOLO by Ultralytics

    2023

  20. [28]

    Jocher, G.; et al. 2023. YOLOv5 by Ultralytics. 2020. URL https://github. com/ultralytics/yolov5

    2023

  21. [29]

    Krause, J.; Deng, J.; Stark, M.; and Fei-Fei, L. 2013. Collecting a large-scale dataset of fine-grained cars

    2013

  22. [30]

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

    2009

  23. [31]

    Kumar, A.; Raghunathan, A.; Jones, R.; Ma, T.; and Liang, P. 2022. Fine-tuning can distort pretrained features and underperform out-of-distribution. arXiv preprint arXiv:2202.10054

    2022 arXiv

  24. [32]

    Li, J.; Zhou, P.; Xiong, C.; and Hoi, S. C. 2020. Prototypical contrastive learning of unsupervised representations. arXiv preprint arXiv:2005.04966

    2020 arXiv

  25. [33]

    Li, X.; Hu, Z.; Ge, Y.; Shan, Y.; and Duan, L.-Y. 2023. Exploring model transferability through the lens of potential energy. In ICCV

    2023

  26. [34]

    Li, Y.; Jia, X.; Sang, R.; Zhu, Y.; Green, B.; Wang, L.; and Gong, B. 2021. Ranking Neural Checkpoints. In CVPR

    2021

  27. [35]

    Lin, T.-Y.; Maire, M.; Belongie, S.; Hays, J.; Perona, P.; Ramanan, D.; Doll \'a r, P.; and Zitnick, C. L. 2014. Microsoft coco: Common objects in context. In ECCV

    2014

  28. [36]

    Liu, X.; Zheng, Y.; Du, Z.; Ding, M.; Qian, Y.; Yang, Z.; and Tang, J. 2023. GPT understands, too. AI Open

    2023

  29. [37]

    Magonis, D. 2022. valorant Dataset. \ https://universe.roboflow.com/daniels-magonis-0pjzx/valorant-9ufcp . Accessed: 2023-01-27

    2022

  30. [38]

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

    2013 arXiv

  31. [39]

    Nguyen, C.; Hassner, T.; Seeger, M.; and Archambeau, C. 2020. Leep: A new measure to evaluate transferability of learned representations. In ICML

    2020

  32. [40]

    Nilsback, M.-E.; and Zisserman, A. 2008. Automated flower classification over a large number of classes. In 2008 Sixth Indian conference on computer vision, graphics & image processing. IEEE

    2008

  33. [41]

    P \'a ndy, M.; Agostinelli, A.; Uijlings, J.; Ferrari, V.; and Mensink, T. 2022. Transferability estimation using bhattacharyya class separability. In CVPR

    2022

  34. [42]

    M.; Vedaldi, A.; Zisserman, A.; and Jawahar, C

    Parkhi, O. M.; Vedaldi, A.; Zisserman, A.; and Jawahar, C. 2012. Cats and dogs. In CVPR

    2012

  35. [43]

    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. 2021. Learning transferable visual models from natural language supervision. In ICML

    2021

  36. [44]

    Roboflow, T. 2022. Blood Cell Detection Dataset. \ https://universe.roboflow.com/team-roboflow/blood-cell-detection-1ekwu . Accessed: 2023-01-18

    2022

  37. [45]

    Russakovsky, O.; Deng, J.; Su, H.; Krause, J.; Satheesh, S.; Ma, S.; Huang, Z.; Karpathy, A.; Khosla, A.; Bernstein, M.; et al. 2015. Imagenet large scale visual recognition challenge. IJCV, 115: 211--252

    2015

  38. [46]

    Sandler, M.; Howard, A.; Zhu, M.; Zhmoginov, A.; and Chen, L.-C. 2018. Mobilenetv2: Inverted residuals and linear bottlenecks. In CVPR

    2018

  39. [47]

    Shao, W.; Zhao, X.; Ge, Y.; Zhang, Z.; Yang, L.; Wang, X.; Shan, Y.; and Luo, P. 2022. Not all models are equal: Predicting model transferability in a self-challenging fisher space. In ECCV

    2022

  40. [48]

    Szegedy, C.; Liu, W.; Jia, Y.; Sermanet, P.; Reed, S.; Anguelov, D.; Erhan, D.; Vanhoucke, V.; and Rabinovich, A. 2015. Going deeper with convolutions. In CVPR

    2015

  41. [49]

    Szegedy, C.; Vanhoucke, V.; Ioffe, S.; Shlens, J.; and Wojna, Z. 2016. Rethinking the inception architecture for computer vision. In CVPR

    2016

  42. [50]

    Tan, M.; Chen, B.; Pang, R.; Vasudevan, V.; Sandler, M.; Howard, A.; and Le, Q. V. 2019. Mnasnet: Platform-aware neural architecture search for mobile. In CVPR

    2019

  43. [51]

    Tian, Y.; Sun, C.; Poole, B.; Krishnan, D.; Schmid, C.; and Isola, P. 2020. What makes for good views for contrastive learning? In ‌NeurIPS

    2020

  44. [52]

    T.; Nguyen, C

    Tran, A. T.; Nguyen, C. V.; and Hassner, T. 2019. Transferability and hardness of supervised classification tasks. In ICCV

    2019

  45. [53]

    Traore, M. 2022. Forklift Dataset. \ https://universe.roboflow.com/mohamed-traore-2ekkp/forklift-dsitv . Accessed: 2023-01-15

    2022

  46. [54]

    Wang, Z.; Luo, Y.; Zheng, L.; Huang, Z.; and Baktashmotlagh, M. 2023. How far pre-trained models are from neural collapse on the target dataset informs their transferability. In ICCV

    2023

  47. [55]

    Wolf, T.; Debut, L.; Sanh, V.; Chaumond, J.; Delangue, C.; Moi, A.; Cistac, P.; Rault, T.; Louf, R.; Funtowicz, M.; et al. 2019. Huggingface's transformers: State-of-the-art natural language processing. arXiv preprint arXiv:1910.03771

    2019 arXiv

  48. [56]

    X.; and Lin, D

    Wu, Z.; Xiong, Y.; Yu, S. X.; and Lin, D. 2018. Unsupervised feature learning via non-parametric instance discrimination. In CVPR

    2018

  49. [57]

    A.; Oliva, A.; and Torralba, A

    Xiao, J.; Hays, J.; Ehinger, K. A.; Oliva, A.; and Torralba, A. 2010. Sun database: Large-scale scene recognition from abbey to zoo. In 2010 IEEE computer society conference on computer vision and pattern recognition

    2010

  50. [58]

    Xu, H.; and Kang, U. 2023. Fast and accurate transferability measurement by evaluating intra-class feature variance. In ICCV

    2023

  51. [59]

    Yosinski, J.; Clune, J.; Bengio, Y.; and Lipson, H. 2014. How transferable are features in deep neural networks? In ‌NeurIPS

    2014

  52. [60]

    You, K.; Liu, Y.; Wang, J.; and Long, M. 2021. Logme: Practical assessment of pre-trained models for transfer learning. In ICML

    2021

  53. [61]

    , " * write output.state after.block = add.period write newline

    ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all...

  54. [62]

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