Pith. sign in

REVIEW 4 major objections 5 minor 51 references

MetaMixUp: Learning Adaptive Interpolation Policy of MixUp with Meta-Learning

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read MetaMixUp learns each sample's interpolation weight and beats plain MixUp on five benchmarks.

desk verdict A useful per-sample MixUp regularizer with consistent empirical gains, but the meta-validation set may not be held out from training and the theory overreaches; worth a careful peer review, not a desk reject. read the letter →

arxiv 1908.10059 v1 pith:DZANYRDZ submitted 2019-08-27 cs.CV cs.LG

Zhijun Mai , Guosheng Hu , Dexiong Chen , Fumin Shen , Heng Tao Shen This is my paper · ORCID
classification cs.CVcs.LG
keywords MixUpdataaugmentationmeta-learninginterpolationpolicyregularizationsemi-supervisedlearningpseudo-labelinggradientLipschitzconstant
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that the main weakness of MixUp is not mixing itself but the random, dataset-independent choice of interpolation weights, which can place synthetic samples inside regions belonging to other classes and make the network underfit. MetaMixUp treats each sample pair's interpolation weight lambda as a learnable parameter and updates it by gradient descent on a small validation set, using one unrolled step of the main model's training. On supervised image benchmarks, the method reports lower test error than vanilla MixUp and AdaMixUp, and in semi-supervised settings it improves pseudo-label training and MixMatch when substituted for MixUp. The paper also argues that MixUp's empirical risk is a lower bound of the gradient Lipschitz constant, which is why a poorly chosen interpolation policy can degrade rather than regularize.

What carries the argument

The central machinery is a bilevel meta-optimization loop. In the inner level, the network takes one gradient step on samples mixed with the current $\lambda$; in the outer level, the gradient of the validation loss with respect to $\lambda$ flows back through that step, and $\lambda$ is updated and projected into [0,1] with a sigmoid. This makes the interpolation policy continuous and trainable. A supporting theoretical object is Proposition 1, which shows that gradient Lipschitz continuity of the predictor implies an inequality bounding the MixUp interpolation error by $\lambda(1-\lambda)\kappa\|x-x'\|^2/2$, so MixUp loss acts as a lower bound on controlling the gradient's Lipschitz constant.

What would settle it

Train MetaMixUp twice on CIFAR-10: once with the 1000 meta-validation images excluded from the training set and once with them included, matching all other settings. If test error is substantially worse in the excluded case, the reported gains depend on tuning lambda on training data rather than on a genuine generalization signal.

Watch

Extended reading notes

Core claim

MetaMixUp establishes that data-adaptive interpolation policies improve MixUp. Instead of drawing lambda from a Beta distribution, it optimizes a per-sample lambda_i by differentiating a validation loss through a single gradient step of the network trained on mixed examples; the updated lambda is then used to re-mix and update the network. The paper reports test error reductions on all five supervised benchmarks tested, with CIFAR-10 error falling from 4.57% (vanilla MixUp) to 3.12% and ImageNet top-1 error from 55.06% to 47.35% under PreActResNet18. It also shows that the learned policy puts more weight on original examples when mixing would cause manifold intrusion, and that in semi-supervised learning, combining MetaMixUp with an asynchronous pseudo-labeling threshold improves pseudo-label baselines and improves MixMatch when used in its place.

Load-bearing premise

The load-bearing assumption is that the small validation set used to learn lambda measures true generalization; the paper draws it from the original training set and does not state that those images are excluded from training, so the validation signal may be tuning lambda to memorized examples rather than to out-of-sample performance.

Editorial extensions

If this is right

  • If MetaMixUp is right, hand-tuning MixUp's Beta-distribution parameter becomes unnecessary: the method adapts lambda per sample pair and reports gains over every fixed setting tested.
  • Because the policy is learned from validation feedback, it should transfer to new datasets without redesigning the augmentation policy, matching or beating MixUp on each benchmark tested.
  • In semi-supervised learning, replacing MixUp with MetaMixUp inside MixMatch lowers test error, so the learned interpolation policy composes with other SSL techniques.
  • The distribution analysis suggests the method learns to avoid mixing when it would create collisions, providing a concrete mechanism for reducing manifold intrusion.

Reading between the lines

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

  • A direct test: if the validation subset used for lambda is held out from training, the reported gains may shrink; the paper does not state that the meta-validation images are excluded from training, so some of the improvement could come from tuning lambda to samples the network can memorize.
  • The learned per-class lambda distributions could be distilled into a static class-pair schedule, making the benefit available without the two-loop meta-computation.
  • The same validation-gradient mechanism could be applied to other continuous augmentation hyperparameters, such as cutout size or cropping scale.
  • The lower-bound view of MixUp suggests that interpolation policies should depend on pair distance; a testable extension is to weight lambda by feature-space distance instead of per-sample meta-gradients.
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 / 5 minor

Summary. The paper proposes MetaMixUp, a method that learns per-sample MixUp interpolation coefficients by gradient-based meta-learning on a validation set, rather than sampling them from a fixed Beta distribution. It also contributes a theoretical perspective linking MixUp loss to the gradient Lipschitz constant (Proposition 1), and an extension to semi-supervised learning via asynchronous pseudo-labeling (APL). In supervised experiments on MNIST, Fashion-MNIST, SVHN, CIFAR-10/100, and ImageNet, MetaMixUp reports lower test error than vanilla MixUp and AdaMixUp across several architectures; in semi-supervised experiments on CIFAR-10 and SVHN, the method improves pseudo-label baselines and gives small gains when plugged into MixMatch.

Significance. If the reported gains are real, MetaMixUp offers a simple, discriminator-free way to make MixUp data-adaptive, which is valuable for regularization and semi-supervised learning. The empirical results are broad (five supervised benchmarks, multiple architectures, SSL splits) and the SSL results in Table III include 5-run means and deviations. The paper also provides a clean analytic observation (Proposition 1) that is correct as a mathematical statement. However, the significance is tempered by several unresolved issues: the meta-validation set may overlap the training set, the supervised tables lack error bars, the APL threshold formula is inconsistent with the pseudocode, and the algorithmic description of how lambda is optimized is ambiguous. These concerns affect the credibility of the central empirical claim and the reproducibility of the method.

major comments (4)
  1. [IV-B] The meta-validation set is sampled from the original training set, and the paper never states that these images are removed from the training set. Section IV-B reads: 'We randomly sample 1000 images (100 per class) from the original training set to construct our meta validation set.' Since the meta-objective in Eqs. (8)-(9) evaluates the validation loss after an inner update on the training set, any overlap makes the meta-loss an in-sample quantity: the learned lambda may be optimized for memorization of those particular images rather than for generalization. This is a load-bearing ambiguity for the central claim that MetaMixUp learns a policy from held-out validation performance. Please state explicitly that the validation set is disjoint from the training set (and from the SSL unlabeled set), or rerun the experiments with a properly held-out validation set.
  2. [III-B / Algorithm 1] Equations (5)-(10) treat lambda as a persistent optimization variable that is updated by Eq. (10), and Eq. (6) defines lambda* as the minimizer of a bilevel objective. However, Algorithm 1 line 2 randomly initializes lambda at every iteration, and line 6 updates only that newly initialized value. Thus the meta-gradient never accumulates across iterations, and the method as written does not implement the bilevel optimization in Eq. (6). If lambda is meant to persist, line 2 should be moved before the loop; if each batch is meant to start from a random lambda, the paper should explain how a one-step correction from a fresh random base constitutes a learned policy and why Eq. (6) is the correct objective. This discrepancy affects the interpretation of the method and of the lambda distributions shown in Figures 6 and 7.
  3. [IV-C / Tables I and III] Table I reports a single error rate per method and dataset without error bars or a stated number of runs. Several headline improvements are small (e.g., CIFAR-10 3.12% vs 3.52% over AdaMixUp; SVHN 2.96% vs 3.12%), so the differences may not be statistically meaningful. In Table III, the improvements of MixMatch+MetaMixUp over MixMatch are within one standard deviation (e.g., 7.69 +/- 0.29 vs 7.87 +/- 0.36 on CIFAR-10 1K; 3.63 +/- 0.45 vs 3.79 +/- 0.65 on SVHN 500), which does not support the statement that MetaMixUp 'improve[s] the previous state-of-the-art results.' Please provide multiple runs with standard deviations for the supervised results and temper the claims about MixMatch improvements.
  4. [III-C] The APL threshold update formula is mathematically inconsistent. The text states: 'the threshold sigma is decreased by sigma_d every K epochs and is defined at t epoch, by sigma_t = sigma_{t-1} - sigma_d * [t/K] * K.' With sigma_d=0.05 and K=30, at epoch 30 this subtracts 0.05*1*30 = 1.5, making sigma negative, and the subtraction is repeated at every epoch. Algorithm 2 (lines 4-6) instead decrements by sigma_d every K epochs. Please give the correct closed-form update (e.g., sigma_t = sigma_0 - sigma_d * floor(t/K)) and make the text, the equation, and the pseudocode consistent.
minor comments (5)
  1. [III-A] Proposition 1 bounds |f(lambda x + (1-lambda) x') - [lambda f(x) + (1-lambda) f(x')]|, which is not the same as the MixUp loss (a loss on the mixed sample with interpolated labels). The sentence 'controlling the Lipschitz constant of the gradient necessarily requires minimizing MixUp loss' (Section III-A) and the interpretation in Section IV-D2 go beyond what the proposition establishes; please soften or substantiate the claim.
  2. [IV-D2] Figure 2 is described in Section IV-D as using 'ResNet-50 on CIFAR-10,' but Table I does not report any ResNet-50 results; please clarify which architecture produced the training-loss curves in Figure 2.
  3. [Algorithm 1] Algorithm 1 line 6 updates lambda* but does not show the sigmoid projection from Eq. (11); state where the projection is applied and whether the gradient is computed before or after projection.
  4. [IV-C / IV-D] There are several typos and inconsistencies: 'constast' (Section IV), 'rubust' (Section IV-C), 'nework' (Algorithm 1 line 8), and 'dose not' (Section IV-D1), plus inconsistent spelling of 'AdaMixup' vs 'AdaMixUp.'
  5. [Table II] In Table II, the MetaMixUp row reports error rates that match the Wide-ResNet-28-10 results in Table I, but the table does not specify the architecture; please state which architecture is used for the hyperparameter comparison.

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity: the meta-objective tunes lambda on a sampled validation set; the test-set claims are independently measured, though the validation set is drawn from the training split without an explicit holdout.

full rationale

I inspected the claimed derivation chain. The core algorithm (Eqs. 4-11 and Algorithms 1-2) defines lambda as an upper-level variable optimized against a validation loss after one inner gradient step, which is a standard bilevel meta-learning formulation and is not definitionally equivalent to the test-set results reported in Tables I-III. Proposition 1 is a direct mathematical implication (P1 => P2) with a supplied proof; it does not assume the target performance claim. The paper cites prior work by its own authors ([28]-[30]) only in the hashing-related related-work discussion, and those citations are not load-bearing for MetaMixUp's method or results. The one concern is experimental: Section IV-B states that the meta-validation set is sampled from the original training set ('We randomly sample 1000 images (100 per class) from the original training set to construct our meta validation set') and does not explicitly say those images are excluded from training, so the meta-objective may be evaluated on in-sample images. This is a methodological ambiguity and a correctness risk, not a circularity: lambda is still fitted only to a validation objective, and the reported test errors are measured on an independent test split. No equation reduces the prediction to its inputs, and no self-citation is doing load-bearing work. Hence the appropriate finding is no significant circularity.

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

The method relies on standard deep learning assumptions (smoothness of the network) and on a common meta-learning approximation (one-step unrolling). No fundamentally new entities are introduced; the free parameters are mostly scheduler and threshold choices.

free parameters (5)
  • Meta step size alpha = 5.0
    Step size for updating the interpolation policy lambda via gradient descent; chosen by hand, not tuned per dataset.
  • Validation set size = 1000 (SL), 500 (SSL)
    Number of images sampled from the training set for the meta-objective; shown to have modest effect (Fig. 3) but is a user choice.
  • Lambda initialization distribution = unspecified
    Each iteration lambda is randomly initialized; the distribution is not given, yet it affects the resulting effective policy after one gradient step.
  • Pseudo-label threshold sigma0 = 0.95
    Initial confidence threshold for accepting pseudo-labels in SSL; set constant across experiments.
  • Threshold decay sigma_d and period K = sigma_d=0.05, K=30
    Control how quickly the pseudo-label threshold decreases; paper tests sigma_d sensitivity but the formula as written is erroneous.
assumptions (4)
  • domain assumption The predictive function f is differentiable with kappa-Lipschitz gradient.
    Used in Proposition 1 to derive the MixUp bound; a standard smoothness assumption for theoretical analysis, not verified for the actual networks.
  • ad hoc to paper A single unrolled gradient step on theta is sufficient for the meta-gradient w.r.t. lambda to provide a useful signal.
    Algorithm 1 uses one-step approximation (Eq. 7 and 9); this is a standard meta-learning approximation but is not validated or analyzed in the paper.
  • domain assumption The meta-validation set, sampled from the training set, is a reliable proxy for generalization.
    The method optimizes lambda on this set; if the set overlaps with training data, the proxy is biased.
  • ad hoc to paper Sigmoid projection of the unconstrained lambda preserves the optimization direction.
    Eq. 11 projects lambda to [0,1]; the effect of this nonlinearity on the gradient descent optimization is not discussed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MetaMixUp: Learning Adaptive Interpolation Policy of MixUp with Meta-Learning." pith.science (2026). https://pith.science/paper/DZANYRDZ

@misc{pith2026190810059,
  author       = {Pith},
  title        = {Pith review of: MetaMixUp: Learning Adaptive Interpolation Policy of MixUp with Meta-Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/DZANYRDZ}},
  note         = {Machine review of arXiv:1908.10059}
}
read the original abstract

MixUp is an effective data augmentation method to regularize deep neural networks via random linear interpolations between pairs of samples and their labels. It plays an important role in model regularization, semi-supervised learning and domain adaption. However, despite its empirical success, its deficiency of randomly mixing samples has poorly been studied. Since deep networks are capable of memorizing the entire dataset, the corrupted samples generated by vanilla MixUp with a badly chosen interpolation policy will degrade the performance of networks. To overcome the underfitting by corrupted samples, inspired by Meta-learning (learning to learn), we propose a novel technique of learning to mixup in this work, namely, MetaMixUp. Unlike the vanilla MixUp that samples interpolation policy from a predefined distribution, this paper introduces a meta-learning based online optimization approach to dynamically learn the interpolation policy in a data-adaptive way. The validation set performance via meta-learning captures the underfitting issue, which provides more information to refine interpolation policy. Furthermore, we adapt our method for pseudo-label based semisupervised learning (SSL) along with a refined pseudo-labeling strategy. In our experiments, our method achieves better performance than vanilla MixUp and its variants under supervised learning configuration. In particular, extensive experiments show that our MetaMixUp adapted SSL greatly outperforms MixUp and many state-of-the-art methods on CIFAR-10 and SVHN benchmarks under SSL configuration.

Figures

Figures reproduced from arXiv: 1908.10059 by the authors.

Figure 1
Figure 1. Computation Graph of Our MetaMixUp where λ ∈ [0, 1]. Under MixUp setting, x and x 0 in (P2) can represent xi and xj in (1). Noting that the left term of (P2) is equivalent to the empirical risk of MixUp by replacing the `1-loss with a general loss function and the prediction f(x) and f(x 0 ) with their true labels. MixUp loss can be considered as a proxy of the left term. Now we announce the following proposition wh… view at source ↗
Figure 3
Figure 3. Error rate for different validation size of MetaMixUp on CIFAR-10 [PITH_FULL_IMAGE:figures/full_fig_p009_3.png] view at source ↗
Figure 2
Figure 2. The loss of the supervised training with MetaMixUp and MixUp on [PITH_FULL_IMAGE:figures/full_fig_p009_2.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Feature of MixUp (left) and MetaMixUp (right) on MNIST. [PITH_FULL_IMAGE:figures/full_fig_p010_4.png]
Figure 7
Figure 7. Figure 7: The data-driven interpolation policy distribution learned by [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 5
Figure 5. Figure 5: The distance between mixed samples and different labeled original [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Relative frequency of λ generated via MixUp (red) and MetaMixUp (blue) on MNIST. 4) Distribution of learned interpolation coefficient on CIFAR-10 [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 39 canonical work pages

  1. [1]

    Is object localization for free? - weakly-supervised learning with convolutional neural networks,

    M. Oquab, L. Bottou, I. Laptev, and J. Sivic, “Is object localization for free? - weakly-supervised learning with convolutional neural networks,” in CVPR, 2015

    work page 2015

  2. [2]

    WILDCAT: weakly supervised learning of deep convnets for image classification, pointwise localization and segmentation,

    T. Durand, T. Mordan, N. Thome, and M. Cord, “WILDCAT: weakly supervised learning of deep convnets for image classification, pointwise localization and segmentation,” in CVPR, 2017

    work page 2017

  3. [3]

    Virtual adversarial training: a regularization method for supervised and semi-supervised learning,

    T. Miyato, S.-i. Maeda, S. Ishii, and M. Koyama, “Virtual adversarial training: a regularization method for supervised and semi-supervised learning,” IEEE transactions on pattern analysis and machine intelli- gence, 2018

    work page 2018

  4. [4]

    Mean teachers are better role mod- els: Weight-averaged consistency targets improve semi-supervised deep learning results,

    A. Tarvainen and H. Valpola, “Mean teachers are better role mod- els: Weight-averaged consistency targets improve semi-supervised deep learning results,” in NeurIPS, 2017

    work page 2017

  5. [5]

    Intriguing properties of neural networks,

    C. Szegedy, W. Zaremba, I. Sutskever, J. Bruna, D. Erhan, I. J. Goodfellow, and R. Fergus, “Intriguing properties of neural networks,” CoRR, vol. abs/1312.6199, 2013

    arXiv 2013

  6. [6]

    Shake-shake regularization of 3-branch residual networks,

    X. Gastaldi, “Shake-shake regularization of 3-branch residual networks,” in 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Workshop Track Proceedings , 2017

    work page 2017

  7. [7]

    Dropout: a simple way to prevent neural networks from overfitting,

    N. Srivastava, G. E. Hinton, A. Krizhevsky, I. Sutskever, and R. Salakhutdinov, “Dropout: a simple way to prevent neural networks from overfitting,” Journal of Machine Learning Research, vol. 15, no. 1, pp. 1929–1958, 2014

    work page 1929

  8. [8]

    Imagenet classification with deep convolutional neural networks,

    A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep convolutional neural networks,”Commun. ACM, vol. 60, no. 6, pp. 84–90, 2017

    work page 2017

Show all 51 references
  1. [9]

    mixup: Beyond empirical risk minimization,

    H. Zhang, M. Ciss ´e, Y . N. Dauphin, and D. Lopez-Paz, “mixup: Beyond empirical risk minimization,” CoRR, vol. abs/1710.09412, 2017

    2017 arXiv

  2. [10]

    Mixup as locally linear out-of-manifold regularization,

    H. Guo, Y . Mao, and R. Zhang, “Mixup as locally linear out-of-manifold regularization,” CoRR, vol. abs/1809.02499, 2018

    2018 arXiv

  3. [11]

    Model-agnostic meta-learning for fast adaptation of deep networks,

    C. Finn, P. Abbeel, and S. Levine, “Model-agnostic meta-learning for fast adaptation of deep networks,” in Proceedings of the 34th International Conference on Machine Learning-Volume 70 . JMLR. org, 2017

    2017

  4. [12]

    Practical bayesian optimiza- tion of machine learning algorithms,

    J. Snoek, H. Larochelle, and R. P. Adams, “Practical bayesian optimiza- tion of machine learning algorithms,” in NeurIPS, 2012

    2012

  5. [13]

    Learning unsupervised learning rules,

    L. Metz, N. Maheswaranathan, B. Cheung, and J. Sohl-Dickstein, “Learning unsupervised learning rules,” CoRR, vol. abs/1804.00222, 2018

    2018 arXiv

  6. [14]

    Pseudo-label: The simple and efficient semi-supervised learning method for deep neural networks,

    D.-H. Lee, “Pseudo-label: The simple and efficient semi-supervised learning method for deep neural networks,” in Workshop on Challenges in Representation Learning, ICML , 2013

    2013

  7. [15]

    Mixmatch: A holistic approach to semi-supervised learning,

    D. Berthelot, N. Carlini, I. J. Goodfellow, N. Papernot, A. Oliver, and C. Raffel, “Mixmatch: A holistic approach to semi-supervised learning,” CoRR, vol. abs/1905.02249, 2019

    1905 arXiv

  8. [16]

    Keeping the neural networks simple by minimizing the description length of the weights,

    G. E. Hinton and D. van Camp, “Keeping the neural networks simple by minimizing the description length of the weights,” in Proceedings of the Sixth Annual ACM Conference on Computational Learning Theory, COLT 1993, Santa Cruz, CA, USA, July 26-28, 1993. , 1993, pp. 5–13

    1993

  9. [17]

    Adam: A method for stochastic optimiza- tion,

    D. P. Kingma and J. Ba, “Adam: A method for stochastic optimiza- tion,” in 3rd International Conference on Learning Representations, ICLR 2015, San Diego, CA, USA, May 7-9, 2015, Conference Track Proceedings, 2015

    2015

  10. [18]

    Deep residual learning for image recognition,

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

    2016

  11. [19]

    Au- toaugment: Learning augmentation policies from data,

    E. D. Cubuk, B. Zoph, D. Mane, V . Vasudevan, and Q. V . Le, “Au- toaugment: Learning augmentation policies from data,” arXiv preprint arXiv:1805.09501, 2018

    2018 arXiv

  12. [20]

    Between-class learning for image classification,

    Y . Tokozume, Y . Ushiku, and T. Harada, “Between-class learning for image classification,” in CVPR, 2018

    2018

  13. [21]

    Manifold mixup: Encouraging meaningful on-manifold interpolation as a regularizer,

    V . Verma, A. Lamb, C. Beckham, A. C. Courville, I. Mitliagkas, and Y . Bengio, “Manifold mixup: Encouraging meaningful on-manifold interpolation as a regularizer,” CoRR, vol. abs/1806.05236, 2018

    2018 arXiv

  14. [22]

    Learning to learn: Introduction and overview,

    S. Thrun and L. Pratt, “Learning to learn: Introduction and overview,” in Learning to learn . Springer, 1998

    1998

  15. [23]

    Bengio, S

    Y . Bengio, S. Bengio, and J. Cloutier, Learning a synaptic learning rule. Universit´e de Montr ´eal, D ´epartement d’informatique et de recherche op´erationnelle, 1990

    1990

  16. [24]

    Optimization as a model for few-shot learning,

    S. Ravi and H. Larochelle, “Optimization as a model for few-shot learning,” in ICLR, 2017

    2017

  17. [25]

    Prototypical networks for few- shot learning,

    J. Snell, K. Swersky, and R. S. Zemel, “Prototypical networks for few- shot learning,” in Advances in Neural Information Processing Systems 30: Annual Conference on Neural Information Processing Systems 2017, 4-9 December 2017, Long Beach, CA, USA , 2017, pp. 4080–4090

    2017

  18. [26]

    A simple neural attentive meta-learner,

    N. Mishra, M. Rohaninejad, X. Chen, and P. Abbeel, “A simple neural attentive meta-learner,” in 6th International Conference on Learning Representations, ICLR 2018, Vancouver, BC, Canada, April 30 - May 3, 2018, Conference Track Proceedings , 2018

    2018

  19. [27]

    Rapid adapta- tion with conditionally shifted neurons,

    T. Munkhdalai, X. Yuan, S. Mehri, and A. Trischler, “Rapid adapta- tion with conditionally shifted neurons,” in Proceedings of the 35th International Conference on Machine Learning, ICML 2018, Stock- holmsm¨assan, Stockholm, Sweden, July 10-15, 2018 , 2018, pp. 3661– 3670

    2018

  20. [28]

    Approximate least trimmed sum of squares fitting and applications in image analysis,

    F. Shen, C. Shen, A. van den Hengel, and Z. Tang, “Approximate least trimmed sum of squares fitting and applications in image analysis,”IEEE Trans. Image Processing, vol. 22, no. 5, pp. 1836–1847, 2013

    2013

  21. [29]

    Deep asymmetric pairwise hashing,

    F. Shen, X. Gao, L. Liu, Y . Yang, and H. T. Shen, “Deep asymmetric pairwise hashing,” in Proceedings of the 2017 ACM on Multimedia Conference, MM 2017, Mountain View, CA, USA, October 23-27, 2017 , 2017, pp. 1522–1530

    2017

  22. [30]

    Supervised discrete hashing,

    F. Shen, C. Shen, W. Liu, and H. T. Shen, “Supervised discrete hashing,” in IEEE Conference on Computer Vision and Pattern Recognition, CVPR 2015, Boston, MA, USA, June 7-12, 2015 , 2015, pp. 37–45

    2015

  23. [31]

    Learning to reweight examples for robust deep learning,

    M. Ren, W. Zeng, B. Yang, and R. Urtasun, “Learning to reweight examples for robust deep learning,” in ICML, 2018

    2018

  24. [32]

    Making a science of model search: Hyperparameter optimization in hundreds of dimensions for vi- sion architectures,

    J. Bergstra, D. Yamins, and D. D. Cox, “Making a science of model search: Hyperparameter optimization in hundreds of dimensions for vi- sion architectures,” in Proceedings of the 30th International Conference on Machine Learning, ICML 2013, Atlanta, GA, USA, 16-21 June 2013 , ...

    2013

  25. [33]

    Auto-weka: combined selection and hyperparameter optimization of classification algorithms,

    C. Thornton, F. Hutter, H. H. Hoos, and K. Leyton-Brown, “Auto-weka: combined selection and hyperparameter optimization of classification algorithms,” in The 19th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, KDD 2013, Chicago, IL, USA, August 11-1...

    2013

  26. [34]

    Practical bayesian optimiza- tion of machine learning algorithms,

    J. Snoek, H. Larochelle, and R. P. Adams, “Practical bayesian optimiza- tion of machine learning algorithms,” in Advances in Neural Information Processing Systems 25: 26th Annual Conference on Neural Information Processing Systems 2012. Proceedings of a meeting held December 3...

    2012

  27. [35]

    Hyperparameter optimization with approximate gradient,

    F. Pedregosa, “Hyperparameter optimization with approximate gradient,” in Proceedings of the 33nd International Conference on Machine Learn- ing, ICML 2016, New York City, NY, USA, June 19-24, 2016 , 2016, pp. 737–746

    2016

  28. [36]

    Temporal ensembling for semi-supervised learn- ing,

    S. Laine and T. Aila, “Temporal ensembling for semi-supervised learn- ing,” CoRR, vol. abs/1610.02242, 2016

    2016 arXiv

  29. [37]

    Learning classification with unlabeled data,

    V . R. de Sa, “Learning classification with unlabeled data,” in NeurIPS, 1994

    1994

  30. [38]

    Asymmetric tri-training for unsu- pervised domain adaptation,

    K. Saito, Y . Ushiku, and T. Harada, “Asymmetric tri-training for unsu- pervised domain adaptation,” in Proceedings of the 34th International Conference on Machine Learning, ICML 2017, Sydney, NSW, Australia, 6-11 August 2017 , 2017, pp. 2988–2997

    2017

  31. [39]

    Learning semantic rep- resentations for unsupervised domain adaptation,

    S. Xie, Z. Zheng, L. Chen, and C. Chen, “Learning semantic rep- resentations for unsupervised domain adaptation,” in Proceedings of the 35th International Conference on Machine Learning, ICML 2018, Stockholmsm¨assan, Stockholm, Sweden, July 10-15, 2018 , 2018, pp. 5419–5428

    2018

  32. [40]

    Parseval networks: Improving robustness to adversarial examples,

    M. Cisse, P. Bojanowski, E. Grave, Y . Dauphin, and N. Usunier, “Parseval networks: Improving robustness to adversarial examples,” in ICML. JMLR. org, 2017

    2017

  33. [41]

    Lipschitz-margin training: Scal- able certification of perturbation invariance for deep neural networks,

    Y . Tsuzuku, I. Sato, and M. Sugiyama, “Lipschitz-margin training: Scal- able certification of perturbation invariance for deep neural networks,” in NeurIPS, 2018

    2018

  34. [42]

    Progressive neural architecture search,

    C. Liu, B. Zoph, M. Neumann, J. Shlens, W. Hua, L. Li, L. Fei-Fei, A. L. Yuille, J. Huang, and K. Murphy, “Progressive neural architecture search,” in Computer Vision - ECCV 2018 - 15th European Conference, 12 Munich, Germany, September 8-14, 2018, Proceedings, Part I, 2018, pp. 19–35

    2018

  35. [43]

    An overview of bilevel opti- mization,

    B. Colson, P. Marcotte, and G. Savard, “An overview of bilevel opti- mization,” Annals OR, vol. 153, no. 1, pp. 235–256, 2007

    2007

  36. [44]

    Scalable gradient- based tuning of continuous regularization hyperparameters,

    J. Luketina, T. Raiko, M. Berglund, and K. Greff, “Scalable gradient- based tuning of continuous regularization hyperparameters,” in Proceed- ings of the 33nd International Conference on Machine Learning, ICML 2016, New York City, NY, USA, June 19-24, 2016, 2016, pp. 2952–2960

    2016

  37. [45]

    Realistic evaluation of deep semi-supervised learning algorithms,

    A. Oliver, A. Odena, C. A. Raffel, E. D. Cubuk, and I. J. Goodfellow, “Realistic evaluation of deep semi-supervised learning algorithms,” in NeurIPS, 2018

    2018

  38. [46]

    Reading digits in natural images with unsupervised feature learning,

    Y . Netzer, T. Wang, A. Coates, A. Bissacco, B. Wu, and A. Y . Ng, “Reading digits in natural images with unsupervised feature learning,” 2011

    2011

  39. [47]

    Imagenet large scale visual recognition challenge,

    O. Russakovsky, J. Deng, H. Su, J. Krause, S. Satheesh, S. Ma, Z. Huang, A. Karpathy, A. Khosla, M. S. Bernstein, A. C. Berg, and F. Li, “Imagenet large scale visual recognition challenge,” International Journal of Computer Vision , vol. 115, no. 3, pp. 211–252, 2015

    2015

  40. [48]

    Identity mappings in deep residual networks,

    K. He, X. Zhang, S. Ren, and J. Sun, “Identity mappings in deep residual networks,” in ECCV, 2016

    2016

  41. [49]

    SGDR: stochastic gradient descent with warm restarts,

    I. Loshchilov and F. Hutter, “SGDR: stochastic gradient descent with warm restarts,” in ICLR, 2017

    2017

  42. [50]

    Wide residual networks,

    S. Zagoruyko and N. Komodakis, “Wide residual networks,” in BMVC, 2016

    2016

  43. [51]

    Self-ensembling for visual domain adaptation,

    G. French, M. Mackiewicz, and M. H. Fisher, “Self-ensembling for visual domain adaptation,” in 6th International Conference on Learning Representations, ICLR 2018, Vancouver, BC, Canada, April 30 - May 3, 2018, Conference Track Proceedings , 2018

    2018

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.