Pith. sign in

REVIEW 5 major objections 6 minor 54 references

GLOBE: Trajectory-Aligned Gradient Matching with Structured SparseOptimization for Coreset Selection

T0 review · 5 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Coresets selected from multi-checkpoint gradient trajectories beat single-snapshot baselines across six image benchmarks and five architectures.

desk verdict Plausible and novel combination of trajectory matching and structured sparsity, but the evaluation doesn't separate the new mechanism from a distilled-proxy confound, so the claims are not yet supported. read the letter →

arxiv 2608.02690 v1 pith:76TQCKHV submitted 2026-08-03 cs.LG

Hetian Liu , Jin Cui , Mengcheng Shi , Yanbin Hu , Xinyue Long , Boran Zhao , Pengju Pen This is my paper
classification cs.LG
keywords coresetselectiongradientmatchingtrajectoriesstructuredsparsityGroupLASSOElasticNetdata-efficientlearningon-devicetraining
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 introduces GLOBE, a coreset selection method that chooses a small subset of real training images by matching the full dataset's gradient behavior across multiple training checkpoints rather than at a single snapshot. The central claim is that this trajectory-aware selection, combined with sparse weight optimization, consistently improves downstream test accuracy over existing coreset methods, with the largest gains at low retention ratios. A sympathetic reader would care because coreset selection is a practical way to train models under memory and compute constraints, and GLOBE argues that capturing the temporal evolution of gradients is key to making small subsets preserve the full training signal.

What carries the argument

The central object is the gradient trajectory: for each sample and tracked layer, the per-sample gradients are collected from $T$ checkpoints during proxy training and concatenated into one high-dimensional vector. This trajectory encodes how the sample influences optimization across early, middle, and late training stages. The paper's two-level matching objective aligns the first-order mean of these trajectories and the projected uncentered second-order moments, using random projections to keep the second-order computation tractable. The weight optimization is carried by Group LASSO over class-wise $k$-means clusters of penultimate-layer features, plus Elastic Net for sample-level sparsity, solved by a proximal-gradient style algorithm.

What would settle it

A direct test would train a proxy and a target model, extract per-sample gradient trajectories from the same checkpoints for both, and measure the cosine similarity between proxy and target trajectories; if that similarity is low yet GLOBE still wins on downstream accuracy, then the proposed mechanism of faithful proxy-gradient alignment is not the true source of its gains.

Watch

Extended reading notes

Core claim

GLOBE represents each training sample by a gradient trajectory, i.e., the concatenation of its per-layer gradients computed at several checkpoints during proxy training. It then learns a nonnegative weight vector over all samples by jointly minimizing two reconstruction losses: a first-order loss $\|A\omega-b\|_2^2$ that matches the weighted mean trajectory to the full-data mean, and a second-order loss $\|B\omega-c\|_2^2$ that matches the projected uncentered second moments of trajectory distributions. Group LASSO and Elastic Net regularizers, along with a class-balanced Top-$K$ selection step, turn the continuous weights into a discrete coreset of size $K$. The paper reports that this method achieves the highest accuracy in all 18 evaluated settings (six datasets at 10%, 20%, 30% retention) across five evaluation architectures, with the strongest gains over prior methods at the 10% retention level.

Load-bearing premise

The coreset is selected entirely from gradients of a lightweight proxy model, and the paper assumes these proxy gradients faithfully represent the optimization dynamics of the target architectures it evaluates; no experiment measures the gradient similarity between proxy and target.

Editorial extensions

If this is right

  • At 10% retention, GLOBE improves over the strongest baseline by 1.23 points on CIFAR-10, 1.87 points on CINIC-10, and 2.08 points on ImageNet-100, showing that low-budget coreset training can get closer to full-data accuracy.
  • Ablations attribute +1.68 points to using multi-checkpoint trajectories instead of only the final checkpoint, +4.89 points to Group LASSO, and +9 points to Elastic Net, indicating each component contributes materially to the final accuracy.
  • The method shows that distribution matching plus structured sparsity is a viable template for sample selection on datasets with strongly correlated samples, beyond the greedy or pursuit-based schemes used by prior gradient-matching coresets.
  • Class-balanced Top-K selection maintains category coverage, making the coreset usable for many-class datasets such as ImageNet-1K and CIFAR-100, where the per-class budget can be as small as a few dozen images.

Reading between the lines

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

  • The paper does not measure how closely proxy gradients match target-model gradients; an unstated assumption is that the teacher-to-proxy knowledge distillation makes the proxy's trajectory a faithful stand-in, and a testable extension would directly compare proxy and target gradient trajectories at matched checkpoints to verify this.
  • The second-order matching relies on independent random projections per layer; an implicit premise is that the projected uncentered moments preserve enough covariance structure, and one could vary the projection dimension $m$ to see whether downstream accuracy improves monotonically with it.
  • The accuracy gains are largest at the lowest retention ratio, which suggests the dynamic gradient signal matters most when the data budget is extremely tight, and this could motivate adaptive checkpoint schedules optimized specifically for 1%–5% retention budgets.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 6 minor

Summary. The paper proposes GLOBE, a coreset selection method that represents each training sample by its gradient trajectory across multiple checkpoints of a proxy model, matches first-order (mean) and projected second-order statistics of the full-data trajectory distribution, and learns sparse nonnegative sample weights via Group LASSO and Elastic Net. The final coreset is obtained by class-balanced Top-K on the optimized weights. The authors evaluate GLOBE on six image classification benchmarks (CIFAR-10, CINIC-10, SVHN, ImageNet-100, ImageNet-1K, CIFAR-100) at 10%, 20%, and 30% retention ratios and report that it outperforms existing coreset methods in all 18 settings.

Significance. If the empirical claims are fully supported, GLOBE would be a strong new baseline for gradient-based coreset selection, with a clean convex formulation (modulo the budget-constraint issue discussed below), a practical multi-checkpoint trajectory representation, and a teacher-proxy distillation mechanism that addresses the proxy gap. The paper's strengths include the explicit optimization of a sparse weighting problem rather than greedy selection, and the use of second-order moment matching over trajectories. However, the current manuscript does not yet support the headline claim of consistent superiority because of evaluation gaps and a formulation discrepancy.

major comments (5)
  1. [Structured Sparse Optimization, Eq. (18)] Equation (18) solves min_{ω≥0} of the matching loss plus regularizers, but it contains no simplex or budget constraint on ω, despite the abstract and Section 'Structured Sparse Optimization' stating that nonnegative budget constraints are used. Without a constraint such as Σ_i ω_i = K (or ≤ K), the solution's sparsity is determined only by the regularization constants, and the subsequent Top-K selection in Eq. (20) is not the argmin of the optimization problem that was solved. Please either add the budget constraint to the objective or revise the description of the method.
  2. [Table 1] Table 1 reports accuracy averaged over the five evaluation architectures but gives no standard deviations, number of seeds, or per-architecture breakdown. The reported margins over the strongest baseline are modest (1.23–3.93 percentage points), so run-to-run variance could change the ranking. Add error bars and a table (or figure) with per-architecture results to support the claim that GLOBE consistently outperforms existing methods.
  3. [Teacher-proxy alignment; Table 1] The comparison against gradient-based baselines is confounded by the proxy protocol. GLOBE uses the knowledge-distillation objective in Eq. (4), whereas the manuscript does not specify whether CRAIG, GradMatch, and GLISTER used the same proxy architecture, teacher, distillation temperature, or checkpoint schedule. If baselines used undistiiled proxies, the observed gains may reflect the KD alignment rather than trajectory matching or structured sparsity. Please run all gradient-based methods on the same proxy trajectories and report the full protocol.
  4. [Teacher-proxy alignment] The paper acknowledges that the output-space KD in Eq. (4) does not explicitly align intermediate features or gradients, yet it asserts that this provides 'a reliable basis' for constructing gradient trajectories. This assertion is load-bearing because the entire selection depends on proxy gradients. Provide a quantitative measure of proxy–target gradient similarity (e.g., cosine similarity between proxy and target gradients or trajectory agreement) across the five architectures to validate the assumption.
  5. [Table 2] The ablation study does not isolate the second-order matching term (L_sec in Eq. (13)), which is a central component of the multi-order matching claim. Moreover, the final-checkpoint ablation (88.53 vs. 90.21) shows that the trajectory component contributes only 1.68 points in this setting, while removing Elastic Net or Group LASSO produces much larger drops (9 and 4.89 points, respectively). This suggests the regularizers, not the trajectory alignment, drive most of the improvement; ablating first-order vs second-order matching and reporting the result would help disentangle the contributions.
minor comments (6)
  1. [Introduction and 'Multi-Order Distribution Matching'] The second-order term is called 'covariance' in the Introduction, while Eq. (10)-(13) define it as an uncentered second moment; please make the terminology consistent throughout.
  2. [Section 'Experiment and Aanlysis'] The section heading contains a typo; it should read 'Experiment and Analysis'.
  3. [Figure 3(a)] The horizontal axis label reads 'Smaples Within Group'; it should be 'Samples Within Group'.
  4. [References] The reference list contains two entries, Cui et al. 2023a and 2023b, that appear to be the same paper (same title, authors, and venue); please merge or disambiguate them.
  5. [Implementation details] The proximal-gradient solver is described only briefly; please provide the initialization, convergence criterion, and the distribution of the random projection matrices R_j, as well as the values of γ_1, γ_2, T, m, M, τ, and η used in the experiments.
  6. [Table 1 formatting] Table 1's formatting appears to omit entries for the 'Full' and 'GLOBE' columns at the 20% and 30% rows for several datasets; if this is a rendering artifact, please clarify the table layout.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: GLOBE's selection objective is self-contained and evaluated against external benchmarks; minor self-citations are non-load-bearing.

full rationale

The derivation chain begins with per-sample gradient trajectories (Eq. 5) and defines a sparse-weight objective (Eq. 18) whose optimum is used only as a ranking score; the final discrete coreset is obtained by explicit class-balanced Top-K selection (Eq. 20), which the paper acknowledges is not necessarily identical to the optimization support. The matching targets b and c are computed directly from the full dataset, so the objective is self-contained rather than fitted to downstream test accuracy. No equation defines the claimed outcome in terms of its own input, and no fitted parameter is renamed as a prediction. The only self-citations are to related-work baselines or background (e.g., NMS, Zhao et al. 2025, with overlapping author Ren; Cui et al. 2023 on dataset distillation, with overlapping author Cui); these are used as comparisons or context, not as justifications of GLOBE's correctness. The paper's own stated limitation, that teacher-proxy alignment does not explicitly align intermediate features or gradients, is an empirical validity concern about proxy fidelity rather than a circular derivation. The central claim is an empirical comparison against external benchmarks, and the method's components are independently motivated. Therefore no load-bearing circular step is present.

Assumptions & free parameters 9 free parameters · 4 assumptions · 0 invented entities

The method rests on many unstated hyperparameters and an unvalidated proxy-gradient transfer assumption. The most consequential modeling choice is treating uncentered second moments as covariance, which is not justified. No new physical or conceptual entities are introduced.

free parameters (9)
  • lambda_1 (L1 coefficient) = not reported
    Controls sample-level sparsity; sensitivity shown only as coefficient paths in Fig 3(a), no values or grid.
  • lambda_2 (L2 coefficient) = not reported
    Stabilizes correlated weights; sensitivity in Fig 3(b), no values reported.
  • lambda_g (Group LASSO coefficient) = not reported
    Induces group-level sparsity; sensitivity in Fig 3(c), no values.
  • T (number of checkpoints) = not reported
    Number of training checkpoints used to build gradient trajectories; never specified.
  • m (random projection dimension) = not reported
    Dimension of random projections for second-order moments; never stated.
  • M (number of k-means clusters per class) = not reported
    Number of groups for Group LASSO; never stated.
  • tau (distillation temperature) = not reported
    Knowledge distillation temperature in Eq (3); not specified.
  • eta (distillation blend weight) = not reported
    Balance between soft and hard labels in Eq (4); not specified.
  • gamma_1, gamma_2 (matching weights) = not reported; dropped in Eq (18)
    Combined objective Eq (15) includes these weights, but Eq (18) omits them, unclear whether set to 1.
assumptions (4)
  • domain assumption Lightweight proxy gradients after knowledge distillation are faithful to target model gradients.
    The entire selection uses gradients from a lightweight proxy; the paper claims KD reduces discrepancy but provides no experiment verifying gradient similarity.
  • domain assumption Random projections preserve enough second-order structure for distribution matching.
    The second-order loss uses m-dimensional random projections; the paper does not analyze the projection dimension or its effect on matching quality.
  • domain assumption Class-balanced Top-K on optimized weights produces a coreset that preserves training dynamics.
    The continuous weights are converted to a discrete coreset via class-balanced Top-K, but the optimization objective does not include the budget constraint, so the two stages may be misaligned.
  • ad hoc to paper The uncentered second moment serves as the covariance structure for gradient trajectories.
    The introduction claims matching the covariance, but Eq (10)-(13) use uncentered outer products, not centered covariances.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GLOBE: Trajectory-Aligned Gradient Matching with Structured SparseOptimization for Coreset Selection." pith.science (2026). https://pith.science/paper/76TQCKHV

@misc{pith2026260802690,
  author       = {Pith},
  title        = {Pith review of: GLOBE: Trajectory-Aligned Gradient Matching with Structured SparseOptimization for Coreset Selection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/76TQCKHV}},
  note         = {Machine review of arXiv:2608.02690}
}
read the original abstract

On-device training of deep neural networks is fundamentally constrained by the computational and memory costs of large-scale datasets. Coreset selection offers a practical solution by retaining only a compact subset of real training samples. However, existing gradient-based methods commonly rely on gradients computed at a single model snapshot and employ greedy or pursuit-based selection procedures, limiting their ability to capture evolving optimization dynamics and handle strongly correlated samples. We propose GLOBE (Gradient Local-Balanced Extraction), a trajectory-aligned coreset selection framework that formulates sample selection as a globally optimized sparse weighting problem. GLOBE represents each sample by a gradient trajectory constructed across multiple training checkpoints, thereby capturing its influence throughout different stages of optimization. To preserve the training behavior of the full dataset, we introduce a multi-order matching objective that jointly aligns the first-order mean and projected uncentered second-order moments of gradient trajectories. GLOBE further combines Group LASSO, Elastic Net regularization, and nonnegative budget constraints to induce group- and sample-level sparsity while stabilizing the weights of correlated trajectories. Finally, class-balanced Top-K selection maintains adequate category coverage under limited sampling budgets. Experiments across six benchmarks and five evaluation architectures demonstrate that GLOBE consistently outperforms existing coreset selection methods in downstream test accuracy, particularly at low retention ratios. These results highlight the effectiveness of combining dynamic gradient information, multi-order distribution matching, and structured sparsity for data-efficient learning.

Figures

Figures reproduced from arXiv: 2608.02690 by the authors.

Figure 1
Figure 1. Comparison between conventional gradient￾based coreset selection and GLOBE. Conventional meth￾ods select samples using gradients from a single model snap￾shot, whereas GLOBE models multi-checkpoint gradient tra￾jectories, matches their first- and second-order statistics, and applies structured sparse optimization to obtain a compact and representative coreset. the quality of the selected coreset. In this work, we re… view at source ↗
Figure 2
Figure 2. Overview of the GLOBE framework. GLOBE first aligns a lightweight proxy model with the teacher model and collects multiple training checkpoints. It then constructs sample-level gradient trajectories, matches the first- and second-order trajectory statistics of the full dataset, and learns sparse sample weights using Group LASSO and Elastic Net. Finally, class￾balanced Top-K selection produces the coreset. each track… view at source ↗
Figure 3
Figure 3. Hyperparameter sensitivity and weight analysis of GLOBE. (a) Sparsity paths under different values of λ1. (b) Distribution of positive sample weights as λ2 varies. (c) Group coefficient paths controlled by λg. (d) Heatmap of the learned sample weights in the feature space. under investigation removed [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

54 extracted references · 15 canonical work pages

  1. [1]

    European conference on computer vision , pages=

    End-to-end object detection with transformers , author=. European conference on computer vision , pages=. 2020 , organization=

    2020

  2. [2]

    arXiv preprint arXiv:2010.11929 , year=

    An image is worth 16x16 words: Transformers for image recognition at scale , author=. arXiv preprint arXiv:2010.11929 , year=

    arXiv 2010

  3. [3]

    Proceedings of the IEEE/CVF international conference on computer vision , pages=

    Swin transformer: Hierarchical vision transformer using shifted windows , author=. Proceedings of the IEEE/CVF international conference on computer vision , pages=

  4. [4]

    Advances in neural information processing systems , volume=

    Attention is all you need , author=. Advances in neural information processing systems , volume=

  5. [5]

    Advances in neural information processing systems , volume=

    Language models are few-shot learners , author=. Advances in neural information processing systems , volume=

  6. [6]

    Bert: Pre-training of deep bidirectional transformers for language understanding , author=. Proceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human language technologies, volume 1 (long and short papers) , pages=

    2019

  7. [7]

    Journal of Computational physics , volume=

    Physics-informed neural networks: A deep learning framework for solving forward and inverse problems involving nonlinear partial differential equations , author=. Journal of Computational physics , volume=. 2018 , publisher=

    work page 2018

  8. [8]

    Nature machine intelligence , volume=

    Learning nonlinear operators via DeepONet based on the universal approximation theorem of operators , author=. Nature machine intelligence , volume=. 2021 , publisher=

    2021

Show all 54 references
  1. [9]

    European conference on computer vision , pages=

    Microsoft coco: Common objects in context , author=. European conference on computer vision , pages=. 2014 , organization=

    2014

  2. [10]

    International journal of computer vision , volume=

    Imagenet large scale visual recognition challenge , author=. International journal of computer vision , volume=. 2015 , publisher=

    2015

  3. [11]

    IEEE transactions on pattern analysis and machine intelligence , volume=

    Dataset distillation: A comprehensive review , author=. IEEE transactions on pattern analysis and machine intelligence , volume=. 2023 , publisher=

    2023

  4. [12]

    Advances in Neural Information Processing Systems , volume=

    Efficient dataset distillation using random feature approximation , author=. Advances in Neural Information Processing Systems , volume=

  5. [13]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Improved distribution matching for dataset condensation , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  6. [14]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

    Generalizing dataset distillation via deep generative prior , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages=

  7. [15]

    arXiv preprint arXiv:1811.10959 , year=

    Dataset distillation , author=. arXiv preprint arXiv:1811.10959 , year=

    arXiv

  8. [16]

    Algorithmic Learning Theory , pages=

    Submodular combinatorial information measures with applications in machine learning , author=. Algorithmic Learning Theory , pages=. 2021 , organization=

    2021

  9. [17]

    International Conference on Machine Learning , pages=

    Grad-match: Gradient matching based data subset selection for efficient deep model training , author=. International Conference on Machine Learning , pages=. 2021 , organization=

    2021

  10. [18]

    International Conference on Machine Learning , pages=

    Coresets for data-efficient training of machine learning models , author=. International Conference on Machine Learning , pages=. 2020 , organization=

    2020

  11. [19]

    Proceedings of the AAAI conference on artificial intelligence , volume=

    Glister: Generalization based data subset selection for efficient and robust learning , author=. Proceedings of the AAAI conference on artificial intelligence , volume=

  12. [20]

    Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

    Dataset condensation with distribution matching , author=. Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pages=

  13. [21]

    arXiv preprint arXiv:2006.05929 , year=

    Dataset condensation with gradient matching , author=. arXiv preprint arXiv:2006.05929 , year=

    2006 arXiv

  14. [22]

    International Conference on Machine Learning , pages=

    Dataset condensation via efficient synthetic-data parameterization , author=. International Conference on Machine Learning , pages=. 2022 , organization=

    2022

  15. [23]

    International Conference on Machine Learning , pages=

    Scaling up dataset distillation to imagenet-1k with constant memory , author=. International Conference on Machine Learning , pages=. 2023 , organization=

    2023

  16. [24]

    Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

    Dream: Efficient dataset distillation by representative matching , author=. Proceedings of the IEEE/CVF International Conference on Computer Vision , pages=

  17. [25]

    International Conference on Learning Representations , volume=

    Embarrassingly simple dataset distillation , author=. International Conference on Learning Representations , volume=

  18. [26]

    Advances in neural information processing systems , volume=

    Deep learning on a data diet: Finding important examples early in training , author=. Advances in neural information processing systems , volume=

  19. [27]

    arXiv preprint arXiv:1708.00489 , year=

    Active learning for convolutional neural networks: A core-set approach , author=. arXiv preprint arXiv:1708.00489 , year=

    arXiv

  20. [28]

    International conference on machine learning , pages=

    Submodularity in data subset selection and active learning , author=. International conference on machine learning , pages=. 2015 , organization=

    2015

  21. [29]

    International conference on machine learning , pages=

    Deep bayesian active learning with image data , author=. International conference on machine learning , pages=. 2017 , organization=

    2017

  22. [30]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Learning loss for active learning , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  23. [31]

    arXiv preprint arXiv:1802.09841 , year=

    Adversarial active learning for deep networks: a margin based approach , author=. arXiv preprint arXiv:1802.09841 , year=

    arXiv

  24. [32]

    Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing , pages=

    Active learning by acquiring contrastive examples , author=. Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing , pages=

    2021

  25. [33]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Minimizing the accumulated trajectory error to improve dataset distillation , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  26. [34]

    Advances in Neural Information Processing Systems , volume=

    Sequential subset matching for dataset distillation , author=. Advances in Neural Information Processing Systems , volume=

  27. [35]

    Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

    Dataset distillation by matching training trajectories , author=. Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages=

  28. [36]

    2012 IEEE conference on computer vision and pattern recognition , pages=

    See all by looking at a few: Sparse modeling for finding representative objects , author=. 2012 IEEE conference on computer vision and pattern recognition , pages=. 2012 , organization=

    2012

  29. [37]

    Advances in Neural Information Processing Systems , volume=

    Finding exemplars from pairwise dissimilarities via simultaneous sparse recovery , author=. Advances in Neural Information Processing Systems , volume=

  30. [38]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Regression shrinkage and selection via the lasso , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 1996 , publisher=

    1996

  31. [39]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Regularization and variable selection via the elastic net , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2005 , publisher=

    2005

  32. [40]

    Lasso-type recovery of sparse representations for high-dimensional data , volume=

    Meinshausen, Nicolai and Yu, Bin , year=. Lasso-type recovery of sparse representations for high-dimensional data , volume=. The Annals of Statistics , publisher=. doi:10.1214/07-aos582 , number=

    doi

  33. [41]

    Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=

    Model selection and estimation in regression with grouped variables , author=. Journal of the Royal Statistical Society Series B: Statistical Methodology , volume=. 2006 , publisher=

    2006

  34. [42]

    2009 , publisher=

    Learning multiple layers of features from tiny images , author=. 2009 , publisher=

    2009

  35. [43]

    NIPS workshop on deep learning and unsupervised feature learning , volume=

    Reading digits in natural images with unsupervised feature learning , author=. NIPS workshop on deep learning and unsupervised feature learning , volume=. 2011 , organization=

    2011

  36. [44]

    2009 IEEE conference on computer vision and pattern recognition , pages=

    Imagenet: A large-scale hierarchical image database , author=. 2009 IEEE conference on computer vision and pattern recognition , pages=. 2009 , organization=

    2009

  37. [45]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Describing textures in the wild , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  38. [46]

    Proceedings of the IEEE , volume=

    Remote sensing image scene classification: Benchmark and state of the art , author=. Proceedings of the IEEE , volume=. 2017 , publisher=

    2017

  39. [47]

    2025 , journal=

    NMS: Efficient Edge DNN Training via Near-Memory Sampling on Manifolds , author=. 2025 , journal=

    2025

  40. [48]

    arXiv preprint arXiv:1203.3472 , year=

    Super-samples from kernel herding , author=. arXiv preprint arXiv:1203.3472 , year=

    arXiv

  41. [49]

    Advances in neural information processing systems , volume=

    Imagenet classification with deep convolutional neural networks , author=. Advances in neural information processing systems , volume=

  42. [50]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Deep residual learning for image recognition , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  43. [51]

    Proceedings of the European conference on computer vision (ECCV) , pages=

    Shufflenet v2: Practical guidelines for efficient cnn architecture design , author=. Proceedings of the European conference on computer vision (ECCV) , pages=

  44. [52]

    Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

    Mobilenetv2: Inverted residuals and linear bottlenecks , author=. Proceedings of the IEEE conference on computer vision and pattern recognition , pages=

  45. [53]

    Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops , pages=

    Reducing catastrophic forgetting with learning on synthetic data , author=. Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops , pages=

  46. [54]

    arXiv preprint arXiv:1810.03505 , year=

    Cinic-10 is not imagenet or cifar-10 , author=. arXiv preprint arXiv:1810.03505 , year=

    arXiv

Pith tools

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