Pith. sign in

REVIEW 2 major objections 4 minor 68 references

Confidence Regularized Self-Training

T0 review · 2 major / 4 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 This paper proposes confidence-regularized self-training, which softens pseudo-labels and smooths network outputs, and reports improved accuracy over the unregularized baseline on image classification and semantic segmentation benchmarks.

desk verdict A genuinely useful, code-released regularizer for self-training UDA with a clean theoretical framing, but the abstract overclaims what the experiments show. read the letter →

arxiv 1908.09822 v3 pith:3OINIOWT submitted 2019-08-26 cs.CV cs.LGcs.MMcs.RO

Yang Zou , Zhiding Yu , Xiaofeng Liu , B. V. K. Vijaya Kumar , Jinsong Wang This is my paper · ORCID
classification cs.CVcs.LGcs.MMcs.RO
keywords unsuperviseddomainadaptationself-trainingpseudo-labelsconfidenceregularizationlabelmodelsemanticsegmentationsoftmaxtemperature
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

Deep self-training for unsupervised domain adaptation works by having a network label its own most confident target predictions and retraining on those pseudo-labels, but wrong pseudo-labels can be held with overconfident belief and their errors compound. This paper proposes confidence regularized self-training (CRST), which treats pseudo-labels as continuous latent variables and adds a regularizer that keeps either the pseudo-labels or the network outputs from becoming too sharp. Two families are introduced: label regularization, whose LRENT variant produces soft pseudo-labels, and model regularization, whose MRKLD variant smooths network outputs. On image classification and semantic segmentation benchmarks, CRST improves on the non-regularized CBST baseline, with MRKLD plus LRENT raising mean accuracy on VisDA17 from 76.4% to 78.1% and mIoU on GTA5-to-Cityscapes from 45.9 to 47.1. The point is to prevent self-training from locking in confident mistakes while retaining the alignment benefit of entropy minimization.

What carries the argument

The central object is the regularized self-training objective $\mathcal{L}_{CR}(w,\hat{Y}) = \mathcal{L}_{CB}(w,\hat{Y}) + \alpha R_C(w,\hat{Y})$, where $\mathcal{L}_{CB}$ is the continuous class-balanced self-training loss, optimized by alternating pseudo-label generation and network retraining. Two families of regularizers are introduced: label regularizers that act only on the pseudo-labels and model regularizers that act on the network's softmax output. The paper's key identities are the closed-form LRENT pseudo-label $\hat{y}^{(i)}_t \propto (p(i|x_t)/\lambda_i)^{1/\alpha}$, which is exactly softmax with temperature when the class thresholds are equal, and the equivalence of the MRKLD regularizer to training with pseudo-labels uniformly smoothed by $\epsilon = (K\alpha-\alpha)/(K+K\alpha)$. The alternating optimization is shown to be non-increasing in both steps under convexity, and the whole procedure is an instance of regularized classification maximum likelihood solved by classification expectation maximization.

What would settle it

Run CRST-MRKLD and CBST with identical class-balanced thresholds on a target domain whose network outputs are deliberately miscalibrated, such as one with heavy label noise or a domain shift that inflates confidence on a frequent wrong class; if the regularized model consistently underperforms, the assumption that smoothing helps more than it hurts is refuted.

Watch

Extended reading notes

Core claim

The paper's central claim is that the failures of self-training in unsupervised domain adaptation come mainly from overconfident wrong pseudo-labels, and that adding confidence regularization fixes this without giving up entropy minimization. The authors generalize class-balanced self-training (CBST) to continuous pseudo-labels on a probability simplex, then add a regularizer to the joint loss over network weights and pseudo-labels. Label regularization changes pseudo-label generation, producing soft targets; LRENT's soft pseudo-label is shown to be exactly softmax with temperature when class thresholds are equal. Model regularization keeps hard pseudo-labels but penalizes sharp network outputs; MRKLD is shown to be equivalent to training with pseudo-labels uniformly smoothed by an epsilon term. In experiments, MRKLD and the MRKLD+LRENT combination outperform CBST consistently, with the best single regularizer being MRKLD; the paper also proves the alternating scheme converges and is an instance of regularized classification maximum likelihood solved by classification expectation maximization.

Load-bearing premise

The method assumes that a confident softmax prediction is usually correct often enough that softening all selected pseudo-labels and outputs costs less in true positives than it gains in false positives.

Editorial extensions

If this is right

  • Any CBST-style self-training script can adopt confidence regularization by changing only the pseudo-label generation step (LR) or the retraining loss (MR), leaving the alternating schedule untouched.
  • Because MRKLD adds only a simple gradient term, it brings most of the benefit without storing soft labels, which matters for segmentation where dense labels are large.
  • The equivalence of LRENT to softmax with temperature means existing temperature-annealing recipes can be viewed as confidence-regularized self-training with the same convergence guarantee.
  • Since MRKLD is equivalent to uniformly smoothed pseudo-labels, the paper links label smoothing to self-training and predicts that label smoothing should help in UDA whenever pseudo-labels are noisy.
  • Combining LR and MR outperforms either alone in the reported experiments, suggesting the two regularizers address different error sources: the pseudo-label target and the network's output distribution.

Reading between the lines

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

  • On domains where softmax confidence is a weak predictor of correctness, for example after severe miscalibration, the same regularizer could suppress accurate true positives and lower accuracy, since the paper's own measurements show confidence regularization reduces confidence on true positives as well as false positives.
  • One testable extension is to anneal the regularizer weight $\alpha$ across self-training rounds, using stronger smoothing early when pseudo-labels are noisiest and weaker smoothing later.
  • The same formulation transfers naturally to semi-supervised classification and pseudo-label-based object detection, because the mechanism does not depend on domain-specific losses.
  • The observation that MRKLD outperforms MRL2 and MRENT despite having a steeper gradient near probability one suggests the shape of the regularizer near the one-hot boundary, not just its overall strength, determines how well it prevents confident mistakes.
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

2 major / 4 minor

Summary. The paper proposes Confidence Regularized Self-Training (CRST), a general regularized self-training framework for unsupervised domain adaptation. Pseudo-labels are treated as continuous latent variables optimized by alternating optimization, and two families of regularizers are introduced: label regularization (exemplified by LRENT) and model regularization (MRL2, MRENT, MRKLD). The authors provide theoretical connections: CRST is cast as regularized classification maximum likelihood solved by classification expectation maximization; LRENT soft pseudo-labels are shown to be a generalized softmax with temperature; MRKLD is shown equivalent to label smoothing; and the reverse KLD regularizer is shown equivalent to entropy regularization. Experiments are reported on VisDA17, Office-31, GTA5-to-Cityscapes, and SYNTHIA-to-Cityscapes, with code released.

Significance. If the empirical claims are supported, the paper offers a simple and fairly general way to improve hard pseudo-label self-training, with only a small hyperparameter cost and no need for adversarial training. The derivations in Table 1 and Appendix A are correct and checkable: the LRENT KKT solution, the gradient formulas for MRL2/MRENT/MRKLD, and Propositions 3-5 are all mathematically sound as stated. The release of code and the reporting of five-run means and standard deviations for VisDA17 are strengths. However, the paper's central headline claim that 'CRSTs outperform their non-regularized counterpart' is broader than the reported evidence, and the convergence proof in Proposition 2 relies on an unproved monotonicity assumption for a non-convex network objective.

major comments (2)
  1. [Abstract and Section 6.2, Tables 2 and 5] The abstract's claim that 'CRSTs outperform their non-regularized counterpart' is contradicted by the paper's own tables. In Table 2, MRL2 (76.2±1.0) and MRENT (76.2±0.8) have slightly lower mean accuracy than CBST (76.4±0.9) on VisDA17, which Section 6.2 itself acknowledges by saying these variants 'show slightly worse results.' In Table 5, MRL2 (48.7) and LRENT (48.7) also fall below CBST (48.9) in mIoU on SYNTHIA-to-Cityscapes, and in Table 4 LRENT exactly ties CBST at 45.9 mIoU on GTA5-to-Cityscapes with DeepLabv2. The consistently superior variants are MRKLD and MRKLD+LRENT. The abstract should either restrict the claim to those variants or provide a per-variant significance analysis; with standard deviations overlapping (e.g., 76.2±1.0 vs. 76.4±0.9), the slight differences are not established as meaningful. This is a load-bearing mismatch because the abstract is the take-home message and readers may adopt a regularizer that the reported experiments do not show to be beneficial.
  2. [Appendix B.2, Proposition 2] The convergence proof of Proposition 2 is not complete. The proof asserts that 'Step b) is non-increasing' because gradient descent with a proper learning rate monotonically decreases the retraining loss in Eq. (14). However, for a deep network the retraining objective is non-convex in w, and standard gradient descent does not guarantee monotone decrease without additional conditions (e.g., Lipschitz smoothness of the loss combined with a sufficiently small step size, or a line search) that are neither stated nor verified. The proof then concedes that mini-batch SGD, which is what is used in the experiments, 'may not strictly guarantee the monotonic decrease of the loss.' As written, the convergence theorem is therefore not established for the actual algorithm used. Please state precise assumptions on the network and optimizer under which monotone decrease holds, or explicitly weaken the claim to convergence of the ideal alternating scheme with exact gradient descent and specify the missing regularity conditions.
minor comments (4)
  1. [Table 1] In the LRENT row, the pseudo-label solution is written as (p(i|xt)λk)^(1/α), which is ambiguous; based on Appendix A it should be (p(i|xt)/λk)^(1/α). Please add the division slash for clarity.
  2. [Section 6.3, Table 4] In the ResNet-38 rows of Table 4, the MRL2 and MRENT entries for the 'Truck' class are reported as 3.22 and 2.89, which are surprisingly low compared with the CBST baseline's 30.3 and also unlike the corresponding DeepLabv2 rows (30.3 and 30.2). Please verify these numbers or add a footnote explaining the drop.
  3. [Section 6.2, Office-31] The sentence 'MRKLD+LRENT again outperforms single regularizers' should be softened for Office-31: the improvement over LRENT alone is 86.8 vs. 86.6 mean accuracy, and differences of this size are within the reported standard deviations. A significance test or a more cautious phrasing would be appropriate.
  4. [Section 7.1, Table 8] The CTP/CFP ratio analysis is a useful diagnostic, but the text says confidence regularization benefits self-training because the ratio improves for 'almost all' classes. Since CTP itself decreases for true positives, the paper should explicitly acknowledge that the benefit is heuristic and that no theoretical guarantee links this ratio to final accuracy.

Circularity Check

0 steps flagged · score 2.0 of 10

No significant circularity: the theoretical equivalences follow from stated definitions and the central claim is tested against external benchmarks; the only author-overlapping citation (CBST) is a published baseline, not a load-bearing circular premise.

full rationale

The paper's derivation chain starts from a stated loss (Eq. 5) that adds an explicit regularizer to the CBST loss (Eq. 1), and all pseudo-label solvers and gradient updates in Table 1 are derived by writing the relevant optimization and solving it (e.g., LRENT via Lagrangian/KKT conditions in Appendix A). Propositions 3 through 5 are algebraic identities connecting the proposed regularizers to softmax with temperature and label smoothing; they introduce no fitted parameter and do not assume the result they are used to explain. The main empirical claim is checked on VisDA17, Office-31, GTA5-to-Cityscapes and SYNTHIA-to-Cityscapes against a re-implemented CBST baseline and external published methods. The recurrent citation of [68] (the authors' own CBST paper) is a legitimate, published baseline and supplies the fixed p/lambda_k selection policy; it is not an unverified uniqueness theorem or an ansatz imported to force a conclusion. The abstract's plural claim that 'CRSTs outperform their non-regularized counterpart' is broader than what Tables 2 and 4 support (MRL2 and MRENT are 76.2% vs 76.4% on VisDA17, and LRENT ties CBST at 45.9 mIoU on GTA5-to-Cityscapes), but an overclaimed headline is an accuracy or soundness issue, not a circularity. Accordingly, no circular step is identified; the low score reflects only the presence of an author-overlapping baseline citation that is not load-bearing.

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

The central empirical claim depends on the alpha values and the p schedule as tuned hyperparameters. The main domain assumption is that confidence correlates with correctness. The convergence proof relies on a convexity and monotonicity assumption that is not satisfied by the actual non-convex network training. No new physical or conceptual entities are introduced.

free parameters (2)
  • alpha (regularizer weight) = MRL2: 0.025, MRENT: 0.1, MRKLD: 0.1, LRENT: 0.25
    Chosen by hand and unified across all experiments. Sensitivity analysis is shown only for Office-31 W to A, so the values may not be optimal for other tasks.
  • p schedule for pseudo-label selection = p0 = 20%, delta p = 5% per round, 3 rounds
    Portion of target predictions selected per class. Adopted from CBST and fixed across experiments; sensitivity analysis is reported for one task.
assumptions (3)
  • standard math The entropy regularizer makes the pseudo-label subproblem convex, so KKT conditions give the global minimizer.
    Used in Appendix A to derive the LRENT soft pseudo-label solution.
  • domain assumption Softmax output confidence is a reliable proxy for pseudo-label correctness in the target domain.
    The entire self-training selection procedure (Eq. 4) and the motivation for confidence regularization rely on this. The paper tests it only empirically on four benchmarks.
  • ad hoc to paper The regularizer r_c(w, y_t) is convex with respect to w and y_t, and gradient descent with a proper learning rate monotonically decreases the retraining loss.
    Used in the convergence proof of Proposition 2 (Appendix B.2). For neural networks, the loss is non-convex in w, and mono-tonic decrease is not guaranteed; the authors partially acknowledge this.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Confidence Regularized Self-Training." pith.science (2026). https://pith.science/paper/3OINIOWT

@misc{pith2026190809822,
  author       = {Pith},
  title        = {Pith review of: Confidence Regularized Self-Training},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3OINIOWT}},
  note         = {Machine review of arXiv:1908.09822}
}
read the original abstract

Recent advances in domain adaptation show that deep self-training presents a powerful means for unsupervised domain adaptation. These methods often involve an iterative process of predicting on target domain and then taking the confident predictions as pseudo-labels for retraining. However, since pseudo-labels can be noisy, self-training can put overconfident label belief on wrong classes, leading to deviated solutions with propagated errors. To address the problem, we propose a confidence regularized self-training (CRST) framework, formulated as regularized self-training. Our method treats pseudo-labels as continuous latent variables jointly optimized via alternating optimization. We propose two types of confidence regularization: label regularization (LR) and model regularization (MR). CRST-LR generates soft pseudo-labels while CRST-MR encourages the smoothness on network output. Extensive experiments on image classification and semantic segmentation show that CRSTs outperform their non-regularized counterpart with state-of-the-art performance. The code and models of this work are available at https://github.com/yzou2/CRST.

Figures

Figures reproduced from arXiv: 1908.09822 by the authors.

Figure 1
Figure 1. Illustration of proposed confidence regularization. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Adaptation results on GTA5 → Cityscapes. Rows correspond to sample images in Cityscapes. From left to right, columns correspond to original images, ground truth, and predication results of CBST, MRL2, MRENT, MRKLD, LRENT [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 4
Figure 4. Loss curves regularized by different regularizers. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figures from the paper (7 more)
Figure 5
Figure 5. Figure 5: Minimizers of LRENT and MRKLD. ing MRKLD as example: using Lagrangian multiplier, we can prove the closed-form global minimizer for regularized cross-entropy loss as p ∗(k) = (y (k) + α K )/(1 + α), where k = 1, ..., K is class index. With y being one-hot, the global m…
Figure 6
Figure 6. Figure 6: Mean accuracy versus number of epochs [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Feature visualization for target domain of VisDA17. From left to right: Source model, CBST, MRKLD+LRENT. [PITH_FULL_IMAGE:figures/full_fig_p014_7.png]
Figure 8
Figure 8. Figure 8: Confusion matrices with normalization for CBST and CRSTs. [PITH_FULL_IMAGE:figures/full_fig_p014_8.png]
Figure 9
Figure 9. Figure 9: Histograms of softmax probability entries in target domain of GTA5 [PITH_FULL_IMAGE:figures/full_fig_p014_9.png]
Figure 10
Figure 10. Figure 10: Adaptation results on GTA5 → Cityscapes. Rows correspond to sample images in Cityscapes. From top to bottom, rows correspond to original images, ground truth, and predication results of CBST, MRL2, MRENT, MRKLD, LRENT [PITH_FULL_IMAGE:figures/full_fig_p015_10.png]
Figure 11
Figure 11. Figure 11: Adaptation results on GTA5 → Cityscapes. Rows correspond to sample images in Cityscapes. From top to bottom, rows correspond to original images, ground truth, and pseudo-label maps of CBST, MRL2, MRENT, MRKLD, LRENT [PITH_FULL_IMAGE:figures/full_fig_p016_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

68 extracted references · 55 canonical work pages

  1. [1]

    Semi-supervised logistic regression

    Massih-Reza Amini and Patrick Gallinari. Semi-supervised logistic regression. In ECAI, 2002. 2, 4, 11

    work page 2002

  2. [2]

    Label refinery: Improv- ing imagenet classification through label progression

    Hessam Bagherinezhad, Maxwell Horton, Mohammad Rastegari, and Ali Farhadi. Label refinery: Improv- ing imagenet classification through label progression. arXiv:1805.02641, 2018. 2

    arXiv 2018

  3. [3]

    Convex optimiza- tion

    Stephen Boyd and Lieven Vandenberghe. Convex optimiza- tion. Cambridge university press, 2004. 4, 11

    work page 2004

  4. [4]

    Open set domain adaptation for image and action recognition

    Pau Panareda Busto, Ahsan Iqbal, and Juergen Gall. Open set domain adaptation for image and action recognition. IEEE Trans. PAMI, 2018. 1, 2

    work page 2018

  5. [5]

    Progressive feature alignment for unsupervised do- main adaptation

    Chaoqi Chen, Weiping Xie, Wenbing Huang, Yu Rong, Xinghao Ding, Yue Huang, Tingyang Xu, and Junzhou Huang. Progressive feature alignment for unsupervised do- main adaptation. In CVPR, 2019. 1

    work page 2019

  6. [6]

    Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolu- tion, and fully connected crfs

    Liang-Chieh Chen, George Papandreou, Iasonas Kokkinos, Kevin Murphy, and Alan L Yuille. Deeplab: Semantic image segmentation with deep convolutional nets, atrous convolu- tion, and fully connected crfs. IEEE Trans. PAMI, 2018. 5

    work page 2018

  7. [7]

    Mxnet: A flexible and efficient machine learning library for heterogeneous distributed systems.arXiv preprint arXiv:1512.01274, 2015

    Tianqi Chen, Mu Li, Yutian Li, Min Lin, Naiyan Wang, Minjie Wang, Tianjun Xiao, Bing Xu, Chiyuan Zhang, and Zheng Zhang. Mxnet: A flexible and efficient machine learning library for heterogeneous distributed systems.arXiv preprint arXiv:1512.01274, 2015. 5

    arXiv 2015

  8. [8]

    Learning semantic segmentation from synthetic data: A ge- ometrically guided input-output adaptation approach

    Yuhua Chen, Wen Li, Xiaoran Chen, and Luc Van Gool. Learning semantic segmentation from synthetic data: A ge- ometrically guided input-output adaptation approach. In CVPR, 2019. 2

    work page 2019

Show all 68 references
  1. [9]

    Domain adaptive faster r-cnn for object de- tection in the wild

    Yuhua Chen, Wen Li, Christos Sakaridis, Dengxin Dai, and Luc Van Gool. Domain adaptive faster r-cnn for object de- tection in the wild. In CVPR, 2018. 1

    2018

  2. [10]

    No more discrimi- nation: Cross city adaptation of road scene segmenters

    Yi-Hsin Chen, Wei-Yu Chen, Yu-Ting Chen, Bo-Cheng Tsai, Yu-Chiang Frank Wang, and Min Sun. No more discrimi- nation: Cross city adaptation of road scene segmenters. In ICCV, 2017. 1, 2

    2017

  3. [11]

    The cityscapes dataset for semantic urban scene understanding

    Marius Cordts, Mohamed Omran, Sebastian Ramos, Timo Rehfeld, Markus Enzweiler, Rodrigo Benenson, Uwe Franke, Stefan Roth, and Bernt Schiele. The cityscapes dataset for semantic urban scene understanding. In CVPR,

  4. [12]

    Imagenet: A large-scale hierarchical image database

    Jia Deng, Wei Dong, Richard Socher, Li-Jia Li, Kai Li, and Li Fei-Fei. Imagenet: A large-scale hierarchical image database. In 2009 IEEE conference on computer vision and pattern recognition, pages 248–255. Ieee, 2009. 5

    2009

  5. [13]

    Domain styliza- tion: A fast covariance matching framework towards domain adaptation

    Aysegul Dundar, Ming-Yu Liu, Zhiding Yu, Ting-Chun Wang, John Zedlewski, and Jan Kautz. Domain styliza- tion: A fast covariance matching framework towards domain adaptation. IEEE Trans. PAMI, 2020. 2

    2020

  6. [14]

    Self-ensembling for visual domain adaptation

    Geoffrey French, Michal Mackiewicz, and Mark Fisher. Self-ensembling for visual domain adaptation. ICLR, 2018. 2

    2018

  7. [15]

    Unsupervised domain adaptation by backpropagation

    Yaroslav Ganin and Victor Lempitsky. Unsupervised domain adaptation by backpropagation. In ICML, 2015. 2

    2015

  8. [16]

    Domain-adversarial train- ing of neural networks

    Yaroslav Ganin, Evgeniya Ustinova, Hana Ajakan, Pas- cal Germain, Hugo Larochelle, Franc ¸ois Laviolette, Mario Marchand, and Victor Lempitsky. Domain-adversarial train- ing of neural networks. JMLR, 2016. 6

    2016

  9. [17]

    DLOW: Domain flow for adaptation and generalization

    Rui Gong, Wen Li, Yuhua Chen, and Luc Van Gool. DLOW: Domain flow for adaptation and generalization. In CVPR,

  10. [18]

    Deep Learning

    Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep Learning . MIT Press, 2016. http://www. deeplearningbook.org. 2

    2016

  11. [19]

    Semi-supervised learning by entropy minimization

    Yves Grandvalet and Yoshua Bengio. Semi-supervised learning by entropy minimization. In NeurIPS, 2005. 2, 6

    2005

  12. [20]

    Weinberger

    Chuan Guo, Geoff Pleiss, Yu Sun, and Kilian Q. Weinberger. On calibration of modern neural networks. In ICML, 2017. 4

    2017

  13. [21]

    Unsupervised domain adaptation via calibrat- ing uncertainties

    Ligong Han, Yang Zou, Ruijiang Gao, Lezi Wang, and Dim- itris Metaxas. Unsupervised domain adaptation via calibrat- ing uncertainties. In CVPR Workshops, 2019. 2

    2019

  14. [22]

    Deep residual learning for image recognition

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. Deep residual learning for image recognition. In CVPR,

  15. [23]

    Distilling the knowledge in a neural network

    Geoffrey Hinton, Oriol Vinyals, and Jeff Dean. Distilling the knowledge in a neural network. arXiv:1503.02531, 2015. 2, 4

    2015 arXiv

  16. [24]

    Cycada: Cycle-consistent adversarial domain adapta- tion

    Judy Hoffman, Eric Tzeng, Taesung Park, Jun-Yan Zhu, Phillip Isola, Kate Saenko, Alexei A Efros, and Trevor Dar- rell. Cycada: Cycle-consistent adversarial domain adapta- tion. In ICML, 2018. 1, 2, 5, 7

    2018

  17. [25]

    Cross-domain weakly-supervised object de- tection through progressive domain adaptation

    Naoto Inoue, Ryosuke Furuta, Toshihiko Yamasaki, and Kiy- oharu Aizawa. Cross-domain weakly-supervised object de- tection through progressive domain adaptation. In CVPR,

  18. [26]

    Unsupervised visual domain adaptation: A deep max-margin Gaussian process approach

    Minyoung Kim, Pritish Sahu, Behnam Gholami, and Vladimir Pavlovic. Unsupervised visual domain adaptation: A deep max-margin Gaussian process approach. In CVPR,

  19. [27]

    Temporal ensembling for semi- supervised learning

    Samuli Laine and Timo Aila. Temporal ensembling for semi- supervised learning. In ICLR, 2016. 2

    2016

  20. [28]

    Sliced wasserstein discrepancy for unsu- pervised domain adaptation

    Chen-Yu Lee, Tanmay Batra, Mohammad Haris Baig, and Daniel Ulbricht. Sliced wasserstein discrepancy for unsu- pervised domain adaptation. In CVPR, 2019. 2

    2019

  21. [29]

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

    Dong-Hyun Lee. Pseudo-label: The simple and efficient semi-supervised learning method for deep neural networks. In ICML Workshop on Challenges in Representation Learn- ing, 2013. 1, 2

    2013

  22. [30]

    Microsoft COCO: Common objects in context

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Doll´ar, and C Lawrence Zitnick. Microsoft COCO: Common objects in context. In ECCV, 2014. 5

    2014

  23. [31]

    Feature-level franken- stein: Eliminating variations for discriminative recognition

    Xiaofeng Liu, Site Li, Lingsheng Kong, Wanqing Xie, Ping Jia, Jane You, and BVK Kumar. Feature-level franken- stein: Eliminating variations for discriminative recognition. In CVPR, 2019. 2

    2019

  24. [32]

    Xiaofeng Liu, Yang Zou, Tong Che, Peng Ding, Ping Jia, Jane You, and B.V .K. Kumar. Conservative wasserstein training for pose estimation. In ICCV, 2017. 2

    2017

  25. [33]

    Learning transferable features with deep adaptation networks

    Mingsheng Long, Yue Cao, Jianmin Wang, and Michael I Jordan. Learning transferable features with deep adaptation networks. In ICML, 2015. 2, 6

    2015

  26. [34]

    Conditional adversarial domain adapta- tion

    Mingsheng Long, Zhangjie Cao, Jianmin Wang, and Michael I Jordan. Conditional adversarial domain adapta- tion. In NeurIPS, 2018. 1, 2

    2018

  27. [35]

    Unsupervised domain adaptation with residual trans- fer networks

    Mingsheng Long, Han Zhu, Jianmin Wang, and Michael I Jordan. Unsupervised domain adaptation with residual trans- fer networks. In NeurIPS, 2016. 6

    2016

  28. [36]

    Deep transfer learning with joint adaptation net- works

    Mingsheng Long, Han Zhu, Jianmin Wang, and Michael I Jordan. Deep transfer learning with joint adaptation net- works. In ICML, 2017. 6

    2017

  29. [37]

    Smooth neighbors on teacher graphs for semi-supervised learning

    Yucen Luo, Jun Zhu, Mengxi Li, Yong Ren, and Bo Zhang. Smooth neighbors on teacher graphs for semi-supervised learning. In CVPR, 2018. 2

    2018

  30. [38]

    Image to image translation for domain adaptation

    Zak Murez, Soheil Kolouri, David Kriegman, Ravi Ra- mamoorthi, and Kyungnam Kim. Image to image translation for domain adaptation. In CVPR, 2018. 1

    2018

  31. [39]

    Learning with noisy labels

    Nagarajan Natarajan, Inderjit S Dhillon, Pradeep K Raviku- mar, and Ambuj Tewari. Learning with noisy labels. In NeurIPS, 2013. 2

    2013

  32. [40]

    Open set domain adaptation

    Pau Panareda Busto and Juergen Gall. Open set domain adaptation. In ICCV, 2017. 2

    2017

  33. [41]

    Automatic differentiation in pytorch

    Adam Paszke, Sam Gross, Soumith Chintala, Gregory Chanan, Edward Yang, Zachary DeVito, Zeming Lin, Al- ban Desmaison, Luca Antiga, and Adam Lerer. Automatic differentiation in pytorch. 2017. 5

    2017

  34. [42]

    Visda: A synthetic-to-real benchmark for visual do- main adaptation

    Xingchao Peng, Ben Usman, Neela Kaushik, Dequan Wang, Judy Hoffman, Kate Saenko, Xavier Roynard, Jean- Emmanuel Deschaud, Francois Goulette, Tyler L Hayes, et al. Visda: A synthetic-to-real benchmark for visual do- main adaptation. In CVPR Workshops, 2018. 1, 2, 5

    2018

  35. [43]

    Regularizing neural networks by penalizing confident output distributions

    Gabriel Pereyra, George Tucker, Jan Chorowski, Łukasz Kaiser, and Geoffrey Hinton. Regularizing neural networks by penalizing confident output distributions. In ICLR Work- shop, 2017. 2, 13

    2017

  36. [44]

    Unsupervised domain adaptation with similarity learning

    Pedro O Pinheiro. Unsupervised domain adaptation with similarity learning. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , pages 8004– 8013, 2018. 1, 2, 5, 6

    2018

  37. [45]

    Train- ing deep neural networks on noisy labels with bootstrapping

    Scott Reed, Honglak Lee, Dragomir Anguelov, Christian Szegedy, Dumitru Erhan, and Andrew Rabinovich. Train- ing deep neural networks on noisy labels with bootstrapping. In ICLR, 2015. 2

    2015

  38. [46]

    Playing for data: Ground truth from computer games

    Stephan R Richter, Vibhav Vineet, Stefan Roth, and Vladlen Koltun. Playing for data: Ground truth from computer games. In ECCV, 2016. 1, 5

    2016

  39. [47]

    The SYNTHIA dataset: A large collection of synthetic images for semantic segmen- tation of urban scenes

    German Ros, Laura Sellart, Joanna Materzynska, David Vazquez, and Antonio M Lopez. The SYNTHIA dataset: A large collection of synthetic images for semantic segmen- tation of urban scenes. In CVPR, 2016. 5

    2016

  40. [48]

    Adapting visual category models to new domains

    Kate Saenko, Brian Kulis, Mario Fritz, and Trevor Darrell. Adapting visual category models to new domains. In ECCV,

  41. [49]

    Asymmetric tri-training for unsupervised domain adaptation

    Kuniaki Saito, Yoshitaka Ushiku, and Tatsuya Harada. Asymmetric tri-training for unsupervised domain adaptation. In ICML, 2017. 1, 2

    2017

  42. [50]

    Adversarial dropout regularization

    Kuniaki Saito, Yoshitaka Ushiku, Tatsuya Harada, and Kate Saenko. Adversarial dropout regularization. In ICLR, 2018. 1, 2, 6

    2018

  43. [51]

    Maximum classifier discrepancy for unsuper- vised domain adaptation

    Kuniaki Saito, Kohei Watanabe, Yoshitaka Ushiku, and Tat- suya Harada. Maximum classifier discrepancy for unsuper- vised domain adaptation. 2017. 6, 7

    2017

  44. [52]

    Open set domain adaptation by backpropa- gation

    Kuniaki Saito, Shohei Yamamoto, Yoshitaka Ushiku, and Tatsuya Harada. Open set domain adaptation by backpropa- gation. In ECCV, 2018. 2

    2018

  45. [53]

    Generate to adapt: Aligning domains using generative adversarial networks

    Swami Sankaranarayanan, Yogesh Balaji, Carlos D Castillo, and Rama Chellappa. Generate to adapt: Aligning domains using generative adversarial networks. In CVPR, 2018. 1, 2, 5, 6

    2018

  46. [54]

    A dirt-t approach to unsupervised domain adaptation

    Rui Shu, Hung H Bui, Hirokazu Narui, and Stefano Ermon. A dirt-t approach to unsupervised domain adaptation. In ICLR, 2018. 1, 2

    2018

  47. [55]

    Training convolutional networks with noisy labels

    Sainbayar Sukhbaatar, Joan Bruna, Manohar Paluri, Lubomir Bourdev, and Rob Fergus. Training convolutional networks with noisy labels. arXiv:1406.2080, 2014. 2

    2014 arXiv

  48. [56]

    Deep coral: Correlation alignment for deep domain adaptation

    Baochen Sun and Kate Saenko. Deep coral: Correlation alignment for deep domain adaptation. In ECCV, 2016. 2

    2016

  49. [57]

    Rethinking the inception ar- chitecture for computer vision

    Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jon Shlens, and Zbigniew Wojna. Rethinking the inception ar- chitecture for computer vision. In CVPR, 2016. 2, 12

    2016

  50. [58]

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

    Antti Tarvainen and Harri Valpola. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. In NeurIPS, 2017. 2

    2017

  51. [59]

    Self-labeled techniques for semi-supervised learning: taxon- omy, software and empirical study.Knowledge and Informa- tion Systems, 2015

    Isaac Triguero, Salvador Garc ´ıa, and Francisco Herrera. Self-labeled techniques for semi-supervised learning: taxon- omy, software and empirical study.Knowledge and Informa- tion Systems, 2015. 2

    2015

  52. [60]

    Learning to adapt structured output space for semantic seg- mentation

    Yi-Hsuan Tsai, Wei-Chih Hung, Samuel Schulter, Ki- hyuk Sohn, Ming-Hsuan Yang, and Manmohan Chandraker. Learning to adapt structured output space for semantic seg- mentation. In CVPR, 2018. 1, 2, 5, 7

    2018

  53. [61]

    Adversarial discriminative domain adaptation

    Eric Tzeng, Judy Hoffman, Kate Saenko, and Trevor Darrell. Adversarial discriminative domain adaptation. In CVPR,

  54. [62]

    Deep domain confusion: Maximizing for domain invariance

    Eric Tzeng, Judy Hoffman, Ning Zhang, Kate Saenko, and Trevor Darrell. Deep domain confusion: Maximizing for domain invariance. arXiv:1412.3474, 2014. 2

    2014 arXiv

  55. [63]

    Advent: Adversarial entropy mini- mization for domain adaptation in semantic segmentation

    Tuan-Hung Vu, Himalaya Jain, Maxime Bucher, Matthieu Cord, and Patrick P´erez. Advent: Adversarial entropy mini- mization for domain adaptation in semantic segmentation. In CVPR, 2019. 7

    2019

  56. [64]

    Dcan: Dual channel-wise alignment networks for un- supervised scene adaptation

    Zuxuan Wu, Xintong Han, Yen-Liang Lin, Mustafa Gokhan Uzunbas, Tom Goldstein, Ser Nam Lim, and Larry S Davis. Dcan: Dual channel-wise alignment networks for un- supervised scene adaptation. In ECCV, 2018. 2

    2018

  57. [65]

    Unsupervised word sense disambiguation rivaling supervised methods

    David Yarowsky. Unsupervised word sense disambiguation rivaling supervised methods. In ACL, 1995. 2

    1995

  58. [66]

    Zhiding Yu, Weiyang Liu, Yang Zou, Chen Feng, Srikumar Ramalingam, B. V . K. Vijaya Kumar, and Jan Kautz. Simul- taneous edge alignment and learning. In ECCV, 2018. 2

    2018

  59. [67]

    Fully convolutional adaptation networks for semantic segmentation

    Yiheng Zhang, Zhaofan Qiu, Ting Yao, Dong Liu, and Tao Mei. Fully convolutional adaptation networks for semantic segmentation. In CVPR, 2018. 7

    2018

  60. [68]

    expectation

    Yang Zou, Zhiding Yu, B. V . K. Vijaya Kumar, and Jinsong Wang. Unsupervised domain adaptation for semantic seg- mentation via class-balanced self-training. In ECCV, 2018. 1, 2, 3, 5, 6, 7 Appendix In this appendix, we present the additional details and results that are not co...

    2018

Pith tools

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