SEVA: Leveraging Single-Step Ensemble of Vicinal Augmentations for Test-Time Adaptation

Ling-Yu Duan; Yichun Hu; Zixuan Hu

REVIEW 4 major objections 5 minor 2 cited by

SEVA: Leveraging Single-Step Ensemble of Vicinal Augmentations for Test-Time Adaptation

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read SEVA claims that a closed-form Augmented Entropy loss lets one test-time update stand in for many rounds of vicinal augmentation without extra runtime.

desk verdict SEVA looks like a genuinely useful empirical TTA method, but the central derivation of Augmented Entropy as an upper bound is not valid as written and the proof is deferred to a missing appendix. read the letter →

arxiv 2505.04087 v1 pith:WXLEUBJX submitted 2025-05-07 cs.CV

Zixuan Hu , Yichun Hu , Ling-Yu Duan This is my paper · ORCID

classification cs.CV

keywords test-timeadaptationentropyminimizationvicinalaugmentationfeature-spacesampleselectiondistributionshiftImageNet-Conline

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 tries to establish that test-time adaptation can capture the benefit of many rounds of vicinal data augmentation in a single backward pass. It derives a closed-form loss, Augmented Entropy, as an upper bound on the entropy of a feature perturbed by Gaussian vicinal augmentation, and couples it with a sample-selection rule that uses the same loss's boundary. If the derivation is right, the accuracy gains that previously demanded five to seven explicit augmentation rounds would be obtained at the same runtime as ordinary entropy adaptation, with better accuracy on corrupted images. This matters because augmentation has been shown to stabilize adaptation against overfitting and collapse, but its computational cost has made it impractical for real-time test-time adaptation.

What carries the argument

The central object is the Augmented Entropy loss $L_{\mathrm{AE}}$ of Eq. 5, a closed-form upper bound of the entropy of predictions under Gaussian vicinal augmentation $\tilde{z} \sim \mathcal{N}(z, \Sigma)$ in feature space. It replaces the sample average over $N$ augmented copies with expectation-based terms: the robust prediction $\hat{p}_j = \exp(a_j \cdot z + b_j + \tfrac{1}{2} a_j^\top \Sigma a_j) / \sum_k \exp(a_k \cdot z + b_k + \tfrac{1}{2} a_k^\top \Sigma a_k)$ appears as the outer probability, and a second exponent $\tfrac{1}{2}(a_j-a_i)^\top \Sigma (a_j-a_i)$ weights the pairwise log-sum term. This loss carries the argument because it is both the training objective and the source of the selection mask: $R(x) = \mathbf{1}\{L_{\mathrm{AE}}(x) < \mathcal{L}_0\}$. The class-distance weight in $L_{\mathrm{AE}}$ is what makes the selection mechanism more sensitive to confusing samples.

What would settle it

Take a two-class linear classifier with $a_1=1$, $a_2=0$, $b_1=b_2=0$, $z=0$, and $\Sigma=1$: the true $\mathbb{E}[p_1]$ under $\mathcal{N}(0,1)$ is $0.5$, while the robust prediction $\hat{p}_1 = \sigma(0.5) \approx 0.62$, so the claimed identity behind Eq. 5 fails on this example. A more direct check is to sample $10^5$ vicinal augmentations of an ImageNet-C feature and compare their mean entropy against $L_{\mathrm{AE}}$: if $L_{\mathrm{AE}}$ frequently undercuts the sampled mean, it is not the stated upper bound.

Watch

Extended reading notes

Core claim

SEVA claims that optimizing the closed-form loss $L_{\mathrm{AE}}$ in Eq. 5 produces the effect of minimizing the entropy loss under infinitely many vicinal augmentations, so one training step can stand in for multiple rounds of augmentation training. The loss is assembled from a robust prediction that absorbs the Gaussian perturbation into each logit, plus a second covariance term that amplifies the loss when the model is unsure between two classes whose classifier prototypes are far apart. The paper further claims that the boundary rule $L_{\mathrm{AE}} < \mathcal{L}_0$ selects reliable samples more effectively than vanilla entropy, because it specifically excludes samples whose confusion is between distant classes. On ImageNet-C, these two components together give consistent gains over prior methods in imbalanced label shift, mixed domains, and batch size one, without adding forward or backward passes.

Load-bearing premise

The load-bearing premise is that averaging many noisy class scores and then normalizing (the expectation of a ratio) equals normalizing first and then averaging (the ratio of expectations); this equality is false in general, and the paper defers the two inequalities in Eq. 5 to an appendix that is absent from this version.

Editorial extensions

If this is right

One backward pass of Augmented Entropy can substitute for five to seven explicit vicinal augmentation rounds: SEVA reports accuracy above SAR with seven augmentation rounds while using about one-eighth of its runtime.
Because $L_{\mathrm{AE}}$ doubles as the selection boundary, the method simultaneously chooses reliable samples during adaptation, so no extra entropy-thresholding stage is needed.
The efficiency gain is large enough to make augmentation-style test-time adaptation real-time: processing 50,000 Gaussian-corrupted images takes 103 seconds for SEVA versus 563 and 804 seconds for SAR with five and seven augmentation rounds.
The method is architecture-agnostic within the entropy-based framework: it adapts only the affine parameters of group-normalization and layer-normalization layers and improves both ResNet and ViT across all three wild scenarios.

Reading between the lines

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

If the upper-bound identity were repaired, the same one-step construction would extend to any augmentation whose noise has a tractable moment-generating function, not just Gaussian vicinal noise; the paper does not explore this.
A direct comparison between $L_{\mathrm{AE}}$ and the empirical mean entropy of many sampled augmentations on the same batch would separate the contribution of the bound from a generic regularizing effect; the paper reports neither.
The class-distance weight inside $L_{\mathrm{AE}}$ could be reused as a label-free sample-reliability score in other self-training settings, where filtering confusing examples is likewise the bottleneck.

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.

Desk Editor's Note

read the letter

SEVA is a solid empirical TTA paper whose central theoretical claim does not hold up as written. The empirical package is genuinely good: the component ablations (Table 5) show both the new loss and the selection mechanism matter, Figure 4 shows the selection beats SAR's in F1, and Table 4 supports the efficiency claim directly (103s vs 804s for SAR+VA×7). If the loss works as a heuristic, that is still useful for the community.

The genuinely new bits are the Augmented Entropy loss (Eq. 5), its coupling with the boundary-based selection (Eq. 6), and the analysis in Sec. 3.3.3. The Gaussian moment-matching trick is reused from the authors' prior work [22], and the paper re-derives it without acknowledging that. That is a mild novelty concern, not a fatal one, since the upper-bound loss and selection are new.

The real problem is the derivation of Eq. 5. The limiting step identifies E[p_j] with E[u_j]/E[S], which is false in general; the stress-test counterexample (C=2, a1=1, a2=0, z=0, Sigma=1) gives 0.5 vs 0.62. And the two inequalities are deferred to 'Theorem 1 & 2 in the appendix' — but there is no appendix in this version. So the claim that one backward pass of L_AE reproduces the effect of infinitely many vicinal augmentations is not established. The loss could still be a good surrogate, and the empirical results remain credible, but the mechanism is unproven as written.

Missing pieces that should be addressed in revision: CoTTA and other augmentation-based baselines are cited but absent from the tables; code is not released despite the abstract promising it; and the key hyperparameters (lambda=1.5, L0=ln(1000)) are tuned on the same benchmark used for the headline numbers, with a modest robustness interval shown.

This is not a reject. The empirical contribution is real, the runtime improvement is important, and the selection analysis is interesting. But the theory needs repair — either a genuine bound or an explicit statement that L_AE is a heuristic surrogate. I would send it out for review, asking the authors to supply the appendix and fix or reframe Eq. 5.

Referee Report

4 major / 5 minor

Summary. SEVA proposes a test-time adaptation method that replaces entropy minimization with an 'Augmented Entropy' loss L_AE, derived in Sec. 3.3 as a purported upper bound on the entropy loss under infinitely many Gaussian vicinal augmentations of the feature representation. The method also uses a threshold on L_AE to select reliable samples. The paper reports consistent accuracy gains over prior TTA methods on ImageNet-C in three wild scenarios (imbalanced label shift, mixed corruption, limited batch size) for ResNet50-GN and ViT-Base-LN, with runtime comparisons showing that SEVA is much faster than explicit multi-round augmentation.

Significance. The empirical side is strong and well ablated: Tables 1-3 show consistent gains over strong baselines, Table 5 and Fig. 4 isolate the contributions of the loss and selection mechanism, and Table 4 demonstrates a large runtime advantage over explicit augmentation. If the theoretical claim were established, the paper would offer a genuinely efficient way to capture augmentation effects in a single backward pass. As it stands, however, the central derivation in Eq. 5 is invalid and the supporting appendix is absent, so the contribution is presently an empirically effective heuristic with an accompanying selection mechanism rather than a derived upper bound. The experiments are the main strength; the theoretical framing needs substantial repair before the paper can be accepted.

major comments (4)

[3.3.2, Eq. (5)] The derivation of L_AE is not valid as written. The first equality after taking N to infinity replaces lim_{N→∞} (1/N) Σ_i p_θ(z̃_i)_j, which equals E_{z̃∼N(z,Σ)}[p_θ(z̃)_j] = E[u_j/S], with p̂_j = E[u_j]/E[S] from Eq. 3. These are not equal in general: for C=2, a_1=1, a_2=0, b=0, z=0, and Σ=1, E[p_1] = 1/2 by symmetry, while p̂_1 = σ(0.5) ≈ 0.622. Consequently Eq. 5 does not establish that L_AE is an upper bound on L_∞, and the central claim that one step of L_AE integrates multiple rounds of augmentation is unsupported. A genuine bound is needed, or the claim must be weakened to a heuristic surrogate.
[Eq. (5), 'Theorem 1 & 2 in the appendix'] The two inequalities in Eq. 5 are attributed to 'Theorem 1 & 2 in the appendix', but the submitted version contains no appendix. The reader cannot verify these bounds. This is not a formatting matter: those inequalities are load-bearing for the claimed upper-bound property. The authors should either include the appendix with complete statements and proofs or state the bounds and proofs directly in the main text; otherwise the theoretical result is incomplete.
[3.3.3, Eq. (7)] The comparison between L_AE and entropy in Eq. 7 is not exact. Eq. 5 defines L_AE with the outer coefficient p̂_j = E[u_j]/E[S], whereas Eq. 7 writes p_θ(z)_j as the outer factor. Since p̂_j differs from p_θ(z)_j in general, the claim that the only modification is the additional weight term is inaccurate. Please clarify whether Eq. 7 is an approximation or an equality, and if the latter, reconcile the coefficients. The selection argument in Sec. 3.3.3 depends on this comparison.
[3.3.2 and Section 1 contributions] Even if Eq. 5 were repaired, the statement that optimizing L_AE 'achieves the effect of integrating multiple rounds of impactful augmentation operations into one pass' is stronger than what an upper bound on the loss would justify. An upper bound on L_∞ does not imply that a single gradient step on L_AE follows the same optimization trajectory as multiple gradient steps on the augmented losses. The paper should either prove a statement about gradients or updates, or explicitly describe this as a heuristic motivation that is supported by the experiments.

minor comments (5)

[3.3.1, Eq. (3)] The robust prediction in Eq. 3 is closely related to the perturbation-based prediction in the authors' prior work [22]; since [22] shares an author, please state explicitly what is new in Eq. 3 relative to that work.
[4.1.2] The estimation of Σ_T is described only as 'the variance of a small subset (128 samples) of features from the testing data'. Please specify how and when this estimate is updated during online adaptation, as this is needed for reproducibility.
[Figure 5] The caption of Fig. 5 does not state which model, corruption type, or scenario is used for the hyperparameter sweeps; please add this information.
[Table 4 and Sec. 4.5] The text in Sec. 4.5 refers to 'SAR with 7 rounds of vicinal augmentation', while Table 4 reports both 'SAR+VA×5' and 'SAR+VA×7'; please ensure the textual reference explicitly points to the row used for the one-eighth runtime comparison.
[References] Reference [30] contains a typo ('Internetional Conference'); please correct it.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the main derivation is self-contained but mathematically unproven, and the missing appendix / invalid expectation step is a rigor concern, not a circular reduction.

full rationale

The paper's claimed derivation chain does not reduce to its inputs by construction. Eq. 3 defines the robust prediction p_hat_j = E[e^{a_j z_tilde + b_j}] / E[sum_k e^{a_k z_tilde + b_k}] as an explicit modeling choice, with the paper noting 'we choose to calculate the expectation before normalization, instead of after,' and the Gaussian moment identity is derived in the text rather than imported as a black box. Eq. 5 then proposes L_AE as an upper bound of the infinite-augmentation entropy loss, but the proof is not circular: it is incomplete, since the two inequalities are deferred to 'Theorem. 1 & 2 in the appendix' (no appendix is present in this version) and the limiting step identifying E[p_j] with E[u_j]/E[S] is asserted, not derived, and is false in general. Those are correctness and rigor failures, not equivalences between input and output. The selection mechanism in Eq. 6 thresholds the same L_AE, and the claimed filtering advantage in Eq. 7 follows algebraically from the added weight e^{1/2(a_j-a_i)Sigma(a_j-a_i)^ op}; this is a designed property, not a fitted parameter renamed as a prediction. The only author-overlapping citation, [22], is used for the general observation that feature-space perturbations give rich semantic variation and as background; Eq. 3 is re-derived in the text, so the central TTA-specific loss does not rest on an unverified self-citation. The empirical comparisons in Tables 1-5 and Fig. 4 are independent external evidence. Therefore, despite the proof being unsupported, the paper is not circular in the senses enumerated in the review protocol.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim sits on two fitted hyperparameters (lambda, L0), a test-data-derived covariance, four stated mathematical or domain assumptions, and two unverified bounds (an appendix-referenced inequality and a false-in-general limit equality). The absence of invented physical entities is good, but the ad-hoc equality in Eq. 5 is the main liability.

free parameters (3)

lambda (vicinal range coefficient) = 1.5 (default; ablated over 0.5-3.0)
Scales the covariance of the Gaussian vicinal perturbation, Sigma=lambda*SigmaT (Sec. 4.1.2). Fig. 5 shows performance peaking at 1.5 on the ImageNet-C imbalanced label shift setting; this is the value used in all headline results.
L0 (Augmented Entropy selection boundary) = 1.0 x ln(1000) ≈ 6.9
Threshold in Eq. 6 for filtering samples; set to ln(1000), the uniform entropy over ImageNet classes, and ablated over [0.7,1.2] x ln(1000) in Fig. 5. This is benchmark-aware tuning.
SigmaT estimation subset size = 128 test features
The perturbation covariance is the variance of 128 test-sample features (Sec. 4.1.2); no sensitivity analysis for this subset size is reported.

assumptions (5)

standard math Gaussian moment-generating function: E_{z_hat~N(z,Sigma)}[e^{a·z_hat}] = e^{a·z + (1/2)a·Sigma·a}
Used in Eq. 3 and Eq. 5 to obtain the robust prediction p_hat and the final loss bound.
standard math Jensen-type inequalities bounding the expected entropy by a cross-entropy with the robust prediction and by a log-sum-exp of expectations (Theorems 1 and 2 in the appendix)
The two inequalities in Eq. 5 are asserted with references to the appendix, which is not present in this version; their validity cannot be checked from the main text.
domain assumption Vicinal augmentation is modeled as additive Gaussian noise on the classifier's input feature, z_hat ~ N(z, Sigma)
Sec. 3.3.1; this is the definition of vicinal augmentation used in the theory and matches how Sigma is estimated from test features.
domain assumption The classifier is a linear softmax over features: p(z) = softmax(Az + b) (Eq. 2), and only normalization affine parameters are adapted
Eq. 2 and Sec. 4.1.2; the whole derivation depends on the logits being affine in z, so the theory only applies to the classifier head, not to the backbone.
ad hoc to paper The limit of the sample mean of softmax probabilities equals the ratio of expectations E[u_j]/E[Σ_k u_k] (the equality step in Eq. 5)
This asserts E[p_j] = E[e^{a_j·z_hat+b_j}]/E[Σ_k e^{a_k·z_hat+b_k}], which is false in general; a counterexample with C=2, a1=1, a2=0, z=0, Sigma=1 gives E[p_1]=0.5 but p_hat_1≈0.62. The main text provides no proof.

how reviews work

0 comments

Cite this review

Pith. "Pith review of SEVA: Leveraging Single-Step Ensemble of Vicinal Augmentations for Test-Time Adaptation." pith.science (2026). https://pith.science/paper/WXLEUBJX

@misc{pith2026250504087,
  author       = {Pith},
  title        = {Pith review of: SEVA: Leveraging Single-Step Ensemble of Vicinal Augmentations for Test-Time Adaptation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WXLEUBJX}},
  note         = {Machine review of arXiv:2505.04087}
}

read the original abstract

Test-Time adaptation (TTA) aims to enhance model robustness against distribution shifts through rapid model adaptation during inference. While existing TTA methods often rely on entropy-based unsupervised training and achieve promising results, the common practice of a single round of entropy training is typically unable to adequately utilize reliable samples, hindering adaptation efficiency. In this paper, we discover augmentation strategies can effectively unleash the potential of reliable samples, but the rapidly growing computational cost impedes their real-time application. To address this limitation, we propose a novel TTA approach named Single-step Ensemble of Vicinal Augmentations (SEVA), which can take advantage of data augmentations without increasing the computational burden. Specifically, instead of explicitly utilizing the augmentation strategy to generate new data, SEVA develops a theoretical framework to explore the impacts of multiple augmentations on model adaptation and proposes to optimize an upper bound of the entropy loss to integrate the effects of multiple rounds of augmentation training into a single step. Furthermore, we discover and verify that using the upper bound as the loss is more conducive to the selection mechanism, as it can effectively filter out harmful samples that confuse the model. Combining these two key advantages, the proposed efficient loss and a complementary selection strategy can simultaneously boost the potential of reliable samples and meet the stringent time requirements of TTA. The comprehensive experiments on various network architectures across challenging testing scenarios demonstrate impressive performances and the broad adaptability of SEVA. The code will be publicly available.

Figures

Figures reproduced from arXiv: 2505.04087 by the authors.

**Figure 1.** Multimedia devices often capture corrupted samples, causing distribution shifts that impair model capability. To address this, many TTA methods utilize entropy self-training to adapt the model to these shifts and can use augmented data for multiple training to enhance adaptation, attaining improvements but with significantly increased time overhead. Comparatively, our proposed SEVA achieves the benefits of multi-s… view at source ↗

**Figure 2.** Performance comparison is conducted on the snow type of ImageNet-C under different training strategies. (a) & (b) record accuracy during online adaptation and compare SAR [30], vicinal augmentation, repeat training, and our SEVA on both ResNet and ViT. "Vicinal/Repeat n Times" denotes utilizing vicinal augmentation/repeat training n times based on SAR. (c) records the final accuracy and compares SEVA with the direct… view at source ↗

**Figure 3.** Comparison of different approaches for using vicinal augmentation in the TTA scenario: (a) Direct Augmentation: It involves repeatedly sampling augmented features into training from a Gaussian distribution (as the vicinal area). As the number of repetitions 𝑁 increases, augmented sets become richer, but the computational costs also increase linearly and cannot meet the time requirement. (b) Efficient Approach in SEV… view at source ↗

Figures from the paper (1 more)

**Figure 4.** Figure 4: Results for all four categories validate the superiority of our [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

Adaptive Dual Uncertainty Optimization: Boosting Monocular 3D Object Detection under Test-Time Shifts
cs.CV 2025-08 conditional novelty 6.0 of 10

DUO adapts monocular 3D detectors at test time by combining a label-free conjugate focal loss for semantic uncertainty with a semantic-guided normal-field consistency loss for geometric uncertainty.
Beyond Entropy: Region Confidence Proxy for Wild Test-Time Adaptation
cs.CV 2025-05 conditional novelty 6.0 of 10

ReCAP models each test sample's local feature neighborhood as a Gaussian and optimizes closed-form bounds on regional entropy and instability, improving wild test-time adaptation accuracy.

Reference graph

Works this paper leans on

63 extracted references · 33 canonical work pages · cited by 2 Pith papers

[22]

Xiaotong Li, Zixuan Hu, Jun Liu, Yixiao Ge, Yongxing Dai, and Ling-Yu Duan

work page
[1]

Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. 2016. Layer normaliza- tion. arXiv preprint arXiv:1607.06450 (2016)

arXiv 2016
[2]

Alexander Bartler, Andre Bühler, Felix Wiewel, Mario Döbler, and Bin Yang

work page
[3]

Shai Ben-David, John Blitzer, Koby Crammer, Alex Kulesza, Fernando Pereira, and Jennifer Wortman Vaughan. 2010. A theory of learning from different domains. Machine learning 79 (2010), 151–175

2010
[4]

Olivier Chapelle, Jason Weston, Léon Bottou, and Vladimir Vapnik. 2000. Vicinal risk minimization. Advances in neural information processing systems 13 (2000)

work page 2000
[5]

Dian Chen, Dequan Wang, Trevor Darrell, and Sayna Ebrahimi. 2022. Contrastive test-time adaptation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 295–305

2022
[6]

Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. 2009. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition . Ieee, 248–255

2009
[7]

Terrance DeVries and Graham W Taylor. 2017. Improved regularization of convolutional neural networks with cutout. arXiv preprint arXiv:1708.04552 (2017)

arXiv 2017

Show all 63 references

[8]

Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xi- aohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Sylvain Gelly, et al. 2020. An image is worth 16x16 words: Transformers for image recognition at scale. arXiv prepri...

2020 arXiv
[9]

Yaroslav Ganin and Victor Lempitsky. 2015. Unsupervised domain adaptation by backpropagation. In International conference on machine learning . PMLR, 1180– 1189

2015
[10]

Jin Gao, Jialing Zhang, Xihui Liu, Trevor Darrell, Evan Shelhamer, and Dequan Wang. 2022. Back to the Source: Diffusion-Driven Test-Time Adaptation. arXiv preprint arXiv:2207.03442 (2022)

2022 arXiv
[11]

Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. Deep Residual Learning for Image Recognition. InProceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

2016
[12]

Dan Hendrycks, Steven Basart, Norman Mu, Saurav Kadavath, Frank Wang, Evan Dorundo, Rahul Desai, Tyler Zhu, Samyak Parajuli, Mike Guo, et al. 2021. The many faces of robustness: A critical analysis of out-of-distribution generalization. In Proceedings of the IEEE/CVF internati...

2021
[13]

Dan Hendrycks and Thomas Dietterich. 2019. Benchmarking Neural Network Ro- bustness to Common Corruptions and Perturbations. In International Conference on Learning Representations. https://openreview.net/forum?id=HJz6tiCqYm

2019
[14]

Dan Hendrycks, Norman Mu, Ekin Dogus Cubuk, Barret Zoph, Justin Gilmer, and Balaji Lakshminarayanan. 2019. AugMix: A Simple Data Processing Method to Improve Robustness and Uncertainty. In International Conference on Learning Representations

2019
[15]

Junyuan Hong, Lingjuan Lyu, Jiayu Zhou, and Michael Spranger. 2023. Mecta: Memory-economic continual test-time model adaptation. In 2023 International Conference on Learning Representations

2023
[16]

Xuefeng Hu, Gokhan Uzunbas, Sirius Chen, Rui Wang, Ashish Shah, Ram Neva- tia, and Ser-Nam Lim. 2021. Mixnorm: Test-time adaptation through online normalization estimation. arXiv preprint arXiv:2110.11478 (2021)

2021 arXiv
[17]

Zixuan Hu, Xiaotong Li, Shixiang Tang, Jun Liu, Yichun Hu, and Ling-Yu Duan
[18]

Sheng-Wei Huang, Che-Tsung Lin, Shu-Ping Chen, Yen-Yi Wu, Po-Hao Hsu, and Shang-Hong Lai. 2018. Auggan: Cross domain adaptation with gan-based data augmentation. In Proceedings of the European Conference on Computer Vision (ECCV). 718–731

2018
[19]

Hiroshi Inoue. 2018. Data augmentation by pairing samples for images classifica- tion. arXiv preprint arXiv:1801.02929 (2018)

2018 arXiv
[20]

Sergey Ioffe and Christian Szegedy. 2015. Batch normalization: Accelerating deep network training by reducing internal covariate shift. In International conference on machine learning. pmlr, 448–456

2015
[21]

Pang Wei Koh, Shiori Sagawa, Henrik Marklund, Sang Michael Xie, Marvin Zhang, Akshay Balsubramani, Weihua Hu, Michihiro Yasunaga, Richard Lanas Phillips, Irena Gao, et al. 2021. Wilds: A benchmark of in-the-wild distribution shifts. In International conference on machine learn...

2021
[23]

Hyesu Lim, Byeonggeun Kim, Jaegul Choo, and Sungha Choi. 2022. TTN: A Domain-Shift Aware Batch Normalization in Test-Time Adaptation. In The Eleventh International Conference on Learning Representations

2022
[24]

Soon Hoe Lim, N Benjamin Erichson, Liam Hodgkinson, and Michael W Mahoney
[25]

Benjamin Erichson, Francisco Utrera, Winnie Xu, and Michael W

Soon Hoe Lim, N. Benjamin Erichson, Francisco Utrera, Winnie Xu, and Michael W. Mahoney. 2022. Noisy Feature Mixup. In International Conference on Learning Representations

2022
[26]

Yuejiang Liu, Parth Kothari, Bastien Van Delft, Baptiste Bellot-Gurlet, Taylor Mordan, and Alexandre Alahi. 2021. Ttt++: When does self-supervised test-time training fail or thrive? Advances in Neural Information Processing Systems 34 (2021), 21808–21820

2021
[27]

Agnieszka Mikołajczyk and Michał Grochowski. 2018. Data augmentation for improving deep learning in image classification problem. In 2018 international interdisciplinary PhD workshop (IIPhDW) . IEEE, 117–122

2018
[28]

Humza Naveed, Saeed Anwar, Munawar Hayat, Kashif Javed, and Ajmal Mian
[29]

Shuaicheng Niu, Jiaxiang Wu, Yifan Zhang, Yaofo Chen, Shijian Zheng, Peilin Zhao, and Mingkui Tan. 2022. Efficient test-time model adaptation without forgetting. In International conference on machine learning . PMLR, 16888–16905

2022
[30]

Shuaicheng Niu, Jiaxiang Wu, Yifan Zhang, Zhiquan Wen, Yaofo Chen, Peilin Zhao, and Mingkui Tan. 2023. Towards Stable Test-Time Adaptation in Dynamic Wild World. In Internetional Conference on Learning Representations

2023
[31]

Viktor Olsson, Wilhelm Tranheden, Juliano Pinto, and Lennart Svensson. 2021. Classmix: Segmentation-based data augmentation for semi-supervised learning. In Proceedings of the IEEE/CVF winter conference on applications of computer vision . 1369–1378

2021
[32]

Ori Press, Steffen Schneider, Matthias Kümmerer, and Matthias Bethge. 2024. RDumb: A simple approach that questions our progress in continual test-time adaptation. Advances in Neural Information Processing Systems 36 (2024)

2024
[33]

Engineering Applications of Artificial Intelligence 131 (2024), 107791

Survey: Image mixing and deleting for data augmentation. Engineering Applications of Artificial Intelligence 131 (2024), 107791

2024
[34]

Kuniaki Saito, Kohei Watanabe, Yoshitaka Ushiku, and Tatsuya Harada. 2018. Maximum classifier discrepancy for unsupervised domain adaptation. In Proceed- ings of the IEEE conference on computer vision and pattern recognition . 3723–3732

2018
[35]

Steffen Schneider, Evgenia Rusak, Luisa Eck, Oliver Bringmann, Wieland Brendel, and Matthias Bethge. 2020. Improving robustness against common corruptions by covariate shift adaptation. Advances in neural information processing systems 33 (2020), 11539–11551

2020
[36]

Connor Shorten and Taghi M Khoshgoftaar. 2019. A survey on image data augmentation for deep learning. Journal of big data 6, 1 (2019), 1–48

2019
[37]

Junha Song, Jungsoo Lee, In So Kweon, and Sungha Choi. 2023. Ecotta: Memory- efficient continual test-time adaptation via self-distilled regularization. In Pro- ceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 11920–11929

2023
[38]

Prabhu Teja S and François Fleuret. 2021. Test time Adaptation through Pertur- bation Robustness. In NeurIPS 2021 Workshop on Distribution Shifts: Connecting Methods and Applications

2021
[39]

Yu Sun, Xiaolong Wang, Zhuang Liu, John Miller, Alexei Efros, and Moritz Hardt. 2020. Test-time training with self-supervision for generalization under distribution shifts. In International conference on machine learning . PMLR, 9229– 9248

2020
[40]

Ngoc-Trung Tran, Viet-Hung Tran, Ngoc-Bao Nguyen, Trung-Kien Nguyen, and Ngai-Man Cheung. 2021. On data augmentation for gan training. IEEE Transactions on Image Processing 30 (2021), 1882–1897

2021
[41]

Vladimir Vapnik. 1991. Principles of risk minimization for learning theory. Advances in neural information processing systems 4 (1991)

1991
[42]

Vikas Verma, Alex Lamb, Christopher Beckham, Amir Najafi, Ioannis Mitliagkas, David Lopez-Paz, and Yoshua Bengio. 2019. Manifold mixup: Better represen- tations by interpolating hidden states. In International conference on machine learning. PMLR, 6438–6447

2019
[43]

Peng Su, Kun Wang, Xingyu Zeng, Shixiang Tang, Dapeng Chen, Di Qiu, and Xiaogang Wang. 2020. Adapting object detectors with conditional domain nor- malization. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceedings, Part XI 16 . Sp...

2020
[44]

Dequan Wang, Evan Shelhamer, Shaoteng Liu, Bruno Olshausen, and Trevor Darrell. 2020. Tent: Fully Test-Time Adaptation by Entropy Minimization. In International Conference on Learning Representations

2020
[45]

Mei Wang and Weihong Deng. 2018. Deep visual domain adaptation: A survey. Neurocomputing 312 (2018), 135–153

2018
[46]

Qin Wang, Olga Fink, Luc Van Gool, and Dengxin Dai. 2022. Continual test-time domain adaptation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 7201–7211

2022
[47]

Xiaolong Wang, Ross Girshick, Abhinav Gupta, and Kaiming He. 2018. Non-Local Neural Networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR)

2018
[48]

Riccardo Volpi, Hongseok Namkoong, Ozan Sener, John C Duchi, Vittorio Murino, and Silvio Savarese. 2018. Generalizing to unseen domains via adversarial data augmentation. Advances in neural information processing systems 31 (2018)

2018
[49]

Ross Wightman. 2019. PyTorch Image Models. https://github.com/rwightman/ pytorch-image-models. https://doi.org/10.5281/zenodo.4414861

2019 doi
[50]

Yuxin Wu and Kaiming He. 2018. Group normalization. In Proceedings of the European conference on computer vision (ECCV) . 3–19

2018
[51]

Zhiqi Yu, Jingjing Li, Zhekai Du, Fengling Li, Lei Zhu, and Yang Yang. 2023. Noise-robust continual test-time domain adaptation. In Proceedings of the 31st ACM International Conference on Multimedia . 2654–2662

2023
[52]

Longhui Yuan, Binhui Xie, and Shuang Li. 2023. Robust test-time adaptation in dynamic scenarios. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 15922–15932

2023
[53]

Yanshuo Wang, Jie Hong, Ali Cheraghian, Shafin Rahman, David Ahmedt- Aristizabal, Lars Petersson, and Mehrtash Harandi. 2024. Continual test-time domain adaptation via dynamic sample selection. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision...

2024
[54]

Runhao Zeng, Qi Deng, Huixuan Xu, Shuaicheng Niu, and Jian Chen. 2023. Exploring Motion Cues for Video Test-Time Adaptation. In Proceedings of the 31st ACM International Conference on Multimedia . 1840–1850

2023
[55]

Hongyi Zhang, Moustapha Cisse, Yann N Dauphin, and David Lopez-Paz. 2017. mixup: Beyond empirical risk minimization. arXiv preprint arXiv:1710.09412 (2017)

2017 arXiv
[56]

Marvin Zhang, Sergey Levine, and Chelsea Finn. 2022. Memo: Test time robust- ness via adaptation and augmentation. Advances in neural information processing systems 35 (2022), 38629–38642

2022
[57]

Zhun Zhong, Liang Zheng, Guoliang Kang, Shaozi Li, and Yi Yang. 2020. Random erasing data augmentation. In Proceedings of the AAAI conference on artificial intelligence, Vol. 34. 13001–13008

2020
[58]

Sangdoo Yun, Dongyoon Han, Seong Joon Oh, Sanghyuk Chun, Junsuk Choe, and Youngjoon Yoo. 2019. Cutmix: Regularization strategy to train strong classifiers with localizable features. In Proceedings of the IEEE/CVF international conference on computer vision. 6023–6032

2019
[63]

Barret Zoph, Ekin D Cubuk, Golnaz Ghiasi, Tsung-Yi Lin, Jonathon Shlens, and Quoc V Le. 2020. Learning Data Augmentation Strategies for Object Detection. In European Conference on Computer Vision . 566–583

2020
[2021]

Noisy recurrent neural networks.Advances in Neural Information Processing Systems 34 (2021), 5124–5137

2021
[2022]

In International Conference on Artificial Intelligence and Statistics

Mt3: Meta test-time training for self-supervised test-time adaption. In International Conference on Artificial Intelligence and Statistics. PMLR, 3080–3090
[2023]

Modeling Uncertain Feature Representation for Domain Generalization. (2023). arXiv:2301.06442

2023 arXiv
[2024]

In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition

LEAD: Exploring Logit Space Evolution for Model Selection. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 28664– 28673

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [22]

Xiaotong Li, Zixuan Hu, Jun Liu, Yixiao Ge, Yongxing Dai, and Ling-Yu Duan

work page

[2] [1]

Jimmy Lei Ba, Jamie Ryan Kiros, and Geoffrey E Hinton. 2016. Layer normaliza- tion. arXiv preprint arXiv:1607.06450 (2016)

arXiv 2016

[3] [2]

Alexander Bartler, Andre Bühler, Felix Wiewel, Mario Döbler, and Bin Yang

work page

[4] [3]

Shai Ben-David, John Blitzer, Koby Crammer, Alex Kulesza, Fernando Pereira, and Jennifer Wortman Vaughan. 2010. A theory of learning from different domains. Machine learning 79 (2010), 151–175

2010

[5] [4]

Olivier Chapelle, Jason Weston, Léon Bottou, and Vladimir Vapnik. 2000. Vicinal risk minimization. Advances in neural information processing systems 13 (2000)

work page 2000

[6] [5]

Dian Chen, Dequan Wang, Trevor Darrell, and Sayna Ebrahimi. 2022. Contrastive test-time adaptation. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 295–305

2022

[7] [6]

Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. 2009. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition . Ieee, 248–255

2009

[8] [7]

Terrance DeVries and Graham W Taylor. 2017. Improved regularization of convolutional neural networks with cutout. arXiv preprint arXiv:1708.04552 (2017)

arXiv 2017