Pith. sign in

REVIEW 4 major objections 5 minor 104 references

Your Data Is Not Perfect: Towards Cross-Domain Out-of-Distribution Detection in Class-Imbalanced Data

T0 review · 4 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper introduces a realistic OOD detection setting that combines semantic, domain, and class-imbalance gaps, and proposes a prototype-based network that outperforms prior methods on three benchmarks.

desk verdict The CCOD setting is a real gap and the reported gains are plausible, but Eq. (5) as written makes the OOD threshold nonpositive for every sample, so the published method cannot produce the reported HOS scores. read the letter →

arxiv 2412.06284 v3 pith:SEKQSHG7 submitted 2024-12-09 cs.CV

Xiang Fang , Arvind Easwaran , Blaise Genest , Ponnuthurai Nagaratnam Suganthan This is my paper · ORCID
classification cs.CV
keywords out-of-distributiondetectioncross-domainclass-imbalanceddataprototypealignmentadaptivethresholduncertainty-awareclusteringHOSscore
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

This paper argues that real-world out-of-distribution (OOD) detection must handle three gaps at once: the semantic gap between known and unknown classes, the domain gap between training and test data, and the class-imbalance gap between categories with different sizes. It introduces the class-imbalanced cross-domain OOD detection (CCOD) setting, in which a labeled source domain trains a detector and an unlabeled target domain provides both in-distribution and unknown classes with arbitrary class sizes. To solve CCOD, the paper proposes the UASA network, which builds label-driven prototypes from the source classifier, aligns target features to those prototypes, applies sample-wise adaptive entropy thresholds for OOD decisions, and clusters target samples with uncertainty-derived weights. On three benchmarks, UASA reports higher mean HOS scores than previous state-of-the-art methods, including 70.04 versus 66.41 on DomainNet.

What carries the argument

The key machinery is the label-driven prototype: each column of the source classifier's weight matrix serves as a class prototype, giving a bijective label-to-prototype mapping without extra parameters. These prototypes drive three mechanisms: prototype-guided domain alignment, which minimizes the entropy of target-feature similarities to prototypes and memory-bank neighbors; adaptive threshold generation, which computes a class-wise entropy baseline and rescales it into a per-sample OOD threshold; and uncertainty-aware target clustering, which weights pairwise KL-divergence losses by confidence scores so that semantically similar target samples form tight clusters. Together they target the semantic gap, the domain gap, and the class-imbalance gap respectively.

What would settle it

Measure the entropy distributions of ID and OOD samples in a target domain after training UASA, for example the sketch domain on DomainNet; if a substantial fraction of OOD samples have entropy below the threshold assigned by Equation (5), then Equation (6) misclassifies them as ID and the reported HOS advantage would disappear on that task.

Watch

Extended reading notes

Core claim

UASA's central claim is that class-imbalanced cross-domain OOD detection is tractable with a prototype-based alignment network that avoids fixed decision thresholds. The network treats the columns of the source classifier weight matrix as label-driven prototypes, stores L2-normalized target features in a memory bank, and aligns target samples to prototypes and their neighbors by minimizing the entropy of the similarity distribution. For each target sample it generates an adaptive threshold from class-wise entropy statistics, marking a sample as OOD when its own entropy exceeds that threshold. Uncertainty-aware clustering then pulls together target samples that share pseudo-labels, weighting each pair by confidence. The paper reports mean HOS scores of 70.04 on DomainNet, 61.14 on Office-Home, and 52.39 on VisDA-C, exceeding the best prior methods by 3.63, 3.05, and 3.74 points respectively.

Load-bearing premise

The OOD decision rule rests on the assumption that in-distribution samples always have lower prediction entropy than out-of-distribution samples in the target domain; if some unknown-class samples receive overconfident low-entropy predictions, the adaptive threshold will label them as in-distribution.

Editorial extensions

If this is right

  • CCOD becomes a realistic benchmark for OOD detection in multi-domain, class-imbalanced applications such as autonomous driving and medical imaging.
  • Sample-wise adaptive thresholds offer a general replacement for fixed entropy thresholds in other OOD detectors.
  • Prototype-based alignment can be carried out with only the classifier weights, so the source data need not be revisited during target adaptation.
  • The method's success on DomainNet, where many classes are visually close, suggests adaptive thresholds help most in fine-grained OOD settings.

Reading between the lines

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

  • A direct test of the entropy assumption would be to train UASA on a target domain where OOD classes are designed to be confusable with ID classes and check the fraction of OOD samples below their sample-wise thresholds.
  • Because the prototypes are built from classifier weights, a source-free variant could ship only the weight matrix and adapt to a new target domain without any source images, which the paper suggests but does not evaluate.
  • The uncertainty-weighted clustering loss could be transferred to open-set domain adaptation tasks beyond CCOD, such as open-set semantic segmentation.
  • If the entropy assumption degrades under label shift, an alternative could calibrate thresholds using a small validation set of target ID samples.
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 defines a new task, class-imbalanced cross-domain OOD detection (CCOD), in which a labeled source domain and an unlabeled target domain with possibly different class distributions and unknown classes are used for OOD detection. It proposes the UASA network with four modules: label-driven prototype building, prototype-guided domain alignment, adaptive threshold generation, and uncertainty-aware target clustering. The paper reports large improvements over prior open-set domain adaptation and OOD detection methods on DomainNet, Office-Home, and VisDA-C.

Significance. The proposed CCOD setting is a genuinely useful extension of open-set domain adaptation, and the reported gains (e.g., 70.04 vs. 66.41 mean HOS on DomainNet) are substantial. The paper also provides ablations for each module and hyperparameter studies. However, the published decision rule in Eq. (5) appears to be internally inconsistent with the reported performance, and the sigma inconsistency between Section 4 and Table 6 prevents reproduction. These are local and correctable issues rather than conceptual impossibilities, so the contribution is potentially significant if the equations and implementation details are corrected.

major comments (4)
  1. [Section 3.3, Eq. (5)] The threshold formula in Eq. (5) is internally inconsistent with the reported results. Since 0 <= min(T) <= T_i <= max(T), the numerator T_i - min(T) - max(T) is <= -min(T) <= 0, and with alpha > 0 and log K_s > 0 every threshold o_i is nonpositive. Eq. (6) then labels a sample as ID only if Q(p_i) <= o_i, which for nonnegative entropy is possible only for exactly zero entropy; under a standard softmax this is measure-zero. Thus the published rule would mark almost all target samples as OOD and cannot yield the reported HOS values. Please correct the formula, give the exact deployed decision rule, and provide code or pseudocode for the threshold computation.
  2. [Section 4 vs. Table 6] The implementation details state sigma = 0.05 in Eq. (2), but the ablation in Table 6 reports the full model ('Ours') with results identical to the sigma = 1.0 row and identifies sigma = 1.0 as best. This ambiguity affects all reported results because sigma controls the softmax temperature in Eq. (2) and Eq. (3). Please state explicitly which sigma value was used for Tables 1-4 and the ablations, and why Table 6 uses sigma = 1.0.
  3. [Section 3.3, entropy-ordering assumption] The entire ID/OOD decision rule rests on the assumption that 'ID samples always have lower entropy than OOD samples,' but the paper never validates this ordering on the target domains. If hard OOD samples receive overconfident low-entropy predictions, Eq. (5)-(6) cannot separate them from ID samples. Please add an empirical analysis of the entropy distributions for ID vs. OOD target samples (e.g., histograms or the AUROC of entropy as a detector) and discuss the failure mode. Also clarify how the pseudo-labels in H_i are initialized and updated, since Eq. (4)-(6) form a coupled system.
  4. [Eq. (2)] In Eq. (2), the denominator sums over j = 1 to K_t, but the source classifier has only K_s prototypes and source labels range over {1,...,K_s}. Unless K_t is a typo for K_s, or the sum is over the full label set of size K_s + K_t with a different notation, the source classification probability in Eq. (1) is not well-defined. Please fix the indexing and define K_t before its first use.
minor comments (5)
  1. [Section 4, code availability] The text says 'Our codes are available in Github' but no URL is given; please provide a working link or a supplement, as the central method cannot otherwise be verified.
  2. [Section 4, datasets] For DomainNet and Office-Home, no explicit class-imbalance statistics are reported; please state the class-size distribution or the imbalance factor used, since the CCOD setting is defined by class imbalance.
  3. [Table 8] Table 8 appears to duplicate the row label 'A = 2.4K_s' twice; one of the rows is presumably a different cluster count and should be relabeled.
  4. [Section 4.2] The claim that UASA 'outperforms compared methods by 7.07%' in the RE->PR task should state the specific baseline and table entry, since the sentence is ambiguous as written.
  5. [Table 3] The header formatting in Table 3 is difficult to read; please use separate columns for the imbalance factor mu and the source class count K_s.

Circularity Check

1 steps flagged · score 2.0 of 10

No load-bearing circularity: UASA's HOS scores are grounded in external ground-truth benchmarks, and no fit to target labels is disguised as a prediction; the ATG bootstrapping loop (Eqs. 4-7) is self-referential but does not force the reported results. Eq. (5)'s nonpositive thresholds and Eq. (7)'s boundary-attracting loss are internal-consistency defects (correctness risks), not circular steps.

  1. self definitional [Section 3.3, Eqs. (4)-(7) (Adaptive Threshold Generation)]
    "For convenience, Hi denotes the set of target samples that are labeled as class i... We set the threshold oi of the target sample xt_i using its entropy Ti by: oi = α(Ti − min(T ) − max(T ))/(max(T ) − min(T )) log K s (5)... Thus, for the target sample xt_i, its predicted label is: ¯yt_i = arg max_j pt_ij, Q(pt_i) ≤ o_{arg max_j pt_ij}, OOD, otherwise, (6)."

    The threshold o_i in Eq. (5) is a function of T_i, the class-wise mean entropy of the pseudo-labeled set H_i defined in Eq. (4); the pseudo-labels forming H_i are themselves produced by the threshold rule in Eq. (6) (the text notes 'Initially, we do not distinguish OOD samples', so the labeling is bootstrapped). The ID/OOD boundary is thus defined in terms of the model's own entropy outputs on the same data: threshold -> pseudo-labels -> class entropy -> threshold. Eq. (7) then minimizes ||o'_i - Q(p_i)||^2 (zero once within Delta of the threshold), training sample entropies toward these self-computed thresholds, closing the loop.

full rationale

This paper is an empirical OOD-detection/UDA paper evaluated on three external benchmarks with ground-truth target labels; HOS is the harmonic mean of OS* (ID accuracy) and UNK (OOD recall), and baselines are re-implemented from official code and settings. The derivation chain is therefore self-contained against external evidence: no training loss uses target labels, no reported number is a renamed fit, and the central claim (UASA beats DANCE/Ovanet by 3-9 HOS points) is falsifiable by rerunning the stated protocol. Citations to 'Fang et al. 2024a/b' (even if these are the present authors re-citing themselves) appear only in related-work context and are not load-bearing; the entropy-ordering premise ('ID samples always have lower entropy than OOD samples') is explicitly stated and sourced to external works, not to a self-citation or uniqueness theorem. The one genuinely self-referential element is the ATG module: thresholds (Eq. 5) are computed from class-wise mean entropies of pseudo-labeled samples (Eq. 4), whose labels were assigned by the same threshold rule (Eq. 6), and Eq. (7) then minimizes ||o'_i - Q(p_i)||^2, i.e., it trains entropies toward the self-computed thresholds. This is a bootstrap loop rather than a result-forcing definition, because the HOS evaluation is against ground truth; hence a score of 2, not higher. Separately, the manuscript has internal-consistency defects that are correctness risks, not circularity: Eq. (5) yields o_i <= 0 for every sample (0 <= min(T) <= T_i <= max(T) makes the numerator <= -min(T) <= 0), so Eq. (6) as published would label essentially all softmax samples OOD, contradicting the reported HOS ~61 on Office-Home; Eq. (7) as written attracts entropies to the decision boundary, the opposite of the claimed 'separation'; Section 4 sets sigma = 0.05 in Eq. (2) while Table 6 reports the best model at sigma = 1.0; and Eq. (2) sums the source softmax over K^t rather than K^s. These defects need correction or code confirmation before the published decision rule can be verified, but they do not constitute circular derivation under the taxonomy.

Assumptions & free parameters 6 free parameters · 4 assumptions · 0 invented entities

The method relies on several domain assumptions about entropy ordering, prototype transfer, and cluster separability; these are not derived, and the main free parameters (α, σ, λ, Δ, A) are tuned on single tasks. No new entities are postulated.

free parameters (6)
  • α (threshold scale) = 0.15
    Scales the adaptive threshold in Eq. (5); tuned on the P→R task of DomainNet and reported as robust in a narrow range.
  • temperature σ = 0.05 (main text) / 1.0 (best in Table 6)
    Contradictory: implementation details say σ=0.05 in Eq. (2), but the ablation in Table 6 claims best performance at σ=1.0 and the main results match σ=1.0.
  • loss weights λ1, λ2, λ3 = 0.05, 0.1, 0.1
    Balance Llpb, Lpda, Latg, Luc in Eq. (12); chosen on P→R DomainNet; sensitivity shown only for that task.
  • confidence interval Δ = 0.5
    Margin in the entropy separation loss Eq. (7); best value found by sweeping Δ in Table 7 on VisDA-C.
  • cluster count A = 2.5 Ks
    Number of clusters in uncertainty-aware clustering; swept over 2.4Ks-2.7Ks in Table 8 on VisDA-C, best at 2.5Ks.
  • imbalance factor μ = 5, 10, 50, 100 for VisDA-C
    Controls class imbalance in VisDA-C, but the paper does not describe how classes are down-sampled; for DomainNet and Office-Home, no imbalance construction is given.
assumptions (4)
  • domain assumption ID samples have lower predictive entropy than OOD samples in the target domain.
    Stated in Section 3.3 and used to derive the adaptive threshold (Eq. 5), the ID/OOD rule (Eq. 6), and the separation loss (Eq. 7).
  • domain assumption The classifier weight vectors learned on the source domain are valid prototypes that transfer across domains.
    Used in Section 3.1-3.2 for prototype-guided domain alignment; no theoretical transfer guarantee is provided.
  • domain assumption K-means clustering with A>Ks clusters can separate semantically different OOD classes in the target feature space.
    Assumed in Section 3.4 for uncertainty-aware clustering; depends on feature geometry.
  • domain assumption Semantically similar objects from different domains share the same label.
    Stated in the abstract and introduction as the basis for aligning source and target classes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Your Data Is Not Perfect: Towards Cross-Domain Out-of-Distribution Detection in Class-Imbalanced Data." pith.science (2026). https://pith.science/paper/SEKQSHG7

@misc{pith2026241206284,
  author       = {Pith},
  title        = {Pith review of: Your Data Is Not Perfect: Towards Cross-Domain Out-of-Distribution Detection in Class-Imbalanced Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/SEKQSHG7}},
  note         = {Machine review of arXiv:2412.06284}
}
read the original abstract

Previous OOD detection systems only focus on the semantic gap between ID and OOD samples. Besides the semantic gap, we are faced with two additional gaps: the domain gap between source and target domains, and the class-imbalance gap between different classes. In fact, similar objects from different domains should belong to the same class. In this paper, we introduce a realistic yet challenging setting: class-imbalanced cross-domain OOD detection (CCOD), which contains a well-labeled (but usually small) source set for training and conducts OOD detection on an unlabeled (but usually larger) target set for testing. We do not assume that the target domain contains only OOD classes or that it is class-balanced: the distribution among classes of the target dataset need not be the same as the source dataset. To tackle this challenging setting with an OOD detection system, we propose a novel uncertainty-aware adaptive semantic alignment (UASA) network based on a prototype-based alignment strategy. Specifically, we first build label-driven prototypes in the source domain and utilize these prototypes for target classification to close the domain gap. Rather than utilizing fixed thresholds for OOD detection, we generate adaptive sample-wise thresholds to handle the semantic gap. Finally, we conduct uncertainty-aware clustering to group semantically similar target samples to relieve the class-imbalance gap. Extensive experiments on three challenging benchmarks demonstrate that our proposed UASA outperforms state-of-the-art methods by a large margin.

Figures

Figures reproduced from arXiv: 2412.06284 by the authors.

Figure 1
Figure 1. (a) Example of the class-imbalanced cross-domain out-of-distribution detection [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Overview of our proposed UASA system for the proposed CCOD task. [PITH_FULL_IMAGE:figures/full_fig_p007_2.png] view at source ↗
Figure 3
Figure 3. Visualizations for the P→R task on DomainNet. Left and Middle: T-SNE visu￾alizations of “before domain adaptation” (Left) and “after domain adaptation” (Middle), where red triangles denote source “cat” samples and blue circles denote target “cat” sam￾ples. Right: Visualization of our clustering results. We show partial samples from the “Cat” cluster, where the labeled percentage is larger than 85%. We report the pro… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Left and Middle: ablative CCOD performance in terms of loss (Left) and [PITH_FULL_IMAGE:figures/full_fig_p019_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

104 extracted references · 73 canonical work pages

  1. [1]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  2. [2]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  3. [3]

    write newline

    " write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in ":" * " " * FUNCTION f...

  4. [4]

    , author Qin, K

    author Ainam, J.P. , author Qin, K. , author Owusu, J.W. , author Lu, G. , year 2021 . title Unsupervised domain adaptation for person re-identification with iterative soft clustering . journal Knowledge-Based Systems volume 212 , pages 106644

    2021

  5. [5]

    , author Chakraborty, S

    author Atmakuru, A. , author Chakraborty, S. , author Faust, O. , author Salvi, M. , author Barua, P.D. , author Molinari, F. , author Acharya, U. , author Homaira, N. , year 2024 . title Deep learning in radiology for lung cancer diagnostics: A systematic review of classification, segmentation, and predictive modeling techniques . journal Expert Systems ...

    2024

  6. [6]

    , author Meethal, A

    author Belal, A. , author Meethal, A. , author Romero, F.P. , author Pedersoli, M. , author Granger, E. , year 2024 . title Multi-source domain adaptation for object detection with prototype-based mean teacher , in: booktitle Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision , pp. pages 1277--1286

    2024

  7. [7]

    , author Loghmani, M.R

    author Bucci, S. , author Loghmani, M.R. , author Tommasi, T. , year 2020 . title On the effectiveness of image rotation for open set domain adaptation , in: booktitle European Conference on Computer Vision , organization Springer . pp. pages 422--438

    2020

  8. [8]

    , author Kira, Z

    author Chen, M.H. , author Kira, Z. , author AlRegib, G. , author Yoo, J. , author Chen, R. , author Zheng, J. , year 2019 . title Temporal attentive alignment for large-scale video domain adaptation , in: booktitle Proceedings of the IEEE/CVF International Conference on Computer Vision , pp. pages 6321--6330

    2019

Show all 104 references
  1. [9]

    , author Li, B

    author Chen, M.H. , author Li, B. , author Bao, Y. , author AlRegib, G. , author Kira, Z. , year 2020 . title Action segmentation with joint self-supervised temporal domain adaptation , in: booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogni...

    2020

  2. [10]

    , author Choo, J

    author Cho, W. , author Choo, J. , year 2022 . title Towards accurate open-set recognition via background-class regularization , in: booktitle European Conference on Computer Vision , organization Springer . pp. pages 658--674

    2022

  3. [11]

    , author Sharma, G

    author Choi, J. , author Sharma, G. , author Schulter, S. , author Huang, J.B. , year 2020 . title Shuffle and attend: Video domain adaptation , in: booktitle Computer Vision--ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part XII 16 , org...

    2020

  4. [12]

    , author Zhao, H

    author Tachet des Combes, R. , author Zhao, H. , author Wang, Y.X. , author Gordon, G.J. , year 2020 . title Domain adaptation with conditional distribution matching and generalized label shift . journal Advances in Neural Information Processing Systems volume 33 , pages 19276--19289

    2020

  5. [13]

    , author Zhao, L

    author Cui, H. , author Zhao, L. , author Li, F. , author Zhu, L. , author Han, X. , author Li, J. , year 2024 a. title Effective comparative prototype hashing for unsupervised domain adaptation , in: booktitle Proceedings of the AAAI Conference on Artificial Intelligence , pp...

    2024

  6. [14]

    , author Wang, B

    author Cui, X. , author Wang, B. , author Jiang, S. , author Liu, Z. , author Xu, H. , author Cui, L. , author Li, S. , year 2024 b. title Unified bi-encoder bispace-discriminator disentanglement for cross-domain echocardiography segmentation . journal Knowledge-Based Systems ...

    2024

  7. [15]

    , author Kellenberger, B

    author Damodaran, B.B. , author Kellenberger, B. , author Flamary, R. , author Tuia, D. , author Courty, N. , year 2018 . title Deepjdot: Deep joint distribution optimal transport for unsupervised domain adaptation , in: booktitle Proceedings of the European conference on comp...

    2018

  8. [16]

    , author Pedersen, M

    author De, K. , author Pedersen, M. , year 2021 . title Impact of colour on robustness of deep neural networks , in: booktitle Proceedings of the IEEE/CVF international conference on computer vision , pp. pages 21--30

    2021

  9. [17]

    , author Chang, D

    author Du, R. , author Chang, D. , author Bhunia, A.K. , author Xie, J. , author Ma, Z. , author Song, Y.Z. , author Guo, J. , year 2020 . title Fine-grained visual classification via progressive multi-granularity training of jigsaw patches , in: booktitle European Conference ...

    2020

  10. [18]

    , author Chen, C

    author Fang, Y. , author Chen, C. , author Zhang, W. , author Wu, J. , author Zhang, Z. , author Xie, S. , year 2024 a. title Prototype learning for adversarial domain adaptation . journal Pattern Recognition , pages 110653

    2024

  11. [19]

    , author Yap, P.T

    author Fang, Y. , author Yap, P.T. , author Lin, W. , author Zhu, H. , author Liu, M. , year 2024 b. title Source-free unsupervised domain adaptation: A survey . journal Neural Networks , pages 106230

    2024

  12. [20]

    , author Ren, J

    author Fort, S. , author Ren, J. , author Lakshminarayanan, B. , year 2021 . title Exploring the limits of out-of-distribution detection . journal Advances in Neural Information Processing Systems volume 34 , pages 7068--7081

    2021

  13. [21]

    , author Cao, Z

    author Fu, B. , author Cao, Z. , author Long, M. , author Wang, J. , year 2020 . title Learning to detect open classes for universal domain adaptation , in: booktitle European Conference on Computer Vision , organization Springer . pp. pages 567--583

    2020

  14. [22]

    , author Lempitsky, V

    author Ganin, Y. , author Lempitsky, V. , year 2015 . title Unsupervised domain adaptation by backpropagation , in: booktitle International conference on machine learning , organization PMLR . pp. pages 1180--1189

    2015

  15. [23]

    , author Ma, X

    author Gao, J. , author Ma, X. , author Xu, C. , year 2024 . title Learning transferable conceptual prototypes for interpretable unsupervised domain adaptation . journal IEEE Transactions on Image Processing

    2024

  16. [24]

    , author Zhang, X

    author He, K. , author Zhang, X. , author Ren, S. , author Sun, J. , year 2016 . title Deep residual learning for image recognition , in: booktitle Proceedings of the IEEE conference on computer vision and pattern recognition , pp. pages 770--778

    2016

  17. [25]

    , author Gimpel, K

    author Hendrycks, D. , author Gimpel, K. , year 2016 . title A baseline for detecting misclassified and out-of-distribution examples in neural networks , in: booktitle ICLR

    2016

  18. [26]

    , author Gimpel, K

    author Hendrycks, D. , author Gimpel, K. , year 2017 . title A baseline for detecting misclassified and out-of-distribution examples in neural networks , in: booktitle International Conference on Learning Representations

    2017

  19. [27]

    , author Vinyals, O

    author Hinton, G. , author Vinyals, O. , author Dean, J. , et al., year 2015 . title Distilling the knowledge in a neural network . journal arXiv preprint arXiv:1503.02531 volume 2

    2015 arXiv

  20. [28]

    , author Shen, Y

    author Hsu, Y.C. , author Shen, Y. , author Jin, H. , author Kira, Z. , year 2020 . title Generalized odin: Detecting out-of-distribution image without learning from out-of-distribution data , in: booktitle CVPR , pp. pages 10951--10960

    2020

  21. [29]

    , author Lao, Q

    author Jiang, X. , author Lao, Q. , author Matwin, S. , author Havaei, M. , year 2020 . title Implicit class-conditioned domain alignment for unsupervised domain adaptation , in: booktitle International Conference on Machine Learning , organization PMLR . pp. pages 4816--4827

    2020

  22. [30]

    , author Zhang, J

    author Jiao, W. , author Zhang, J. , author Zhang, C. , year 2024 . title Open-set recognition with long-tail sonar images . journal Expert Systems with Applications volume 249 , pages 123495

    2024

  23. [31]

    , author Jiang, L

    author Kang, G. , author Jiang, L. , author Yang, Y. , author Hauptmann, A.G. , year 2019 . title Contrastive adaptation network for unsupervised domain adaptation , in: booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. pages 4893--4902

    2019

  24. [32]

    , author Izmailov, P

    author Kirichenko, P. , author Izmailov, P. , author Wilson, A.G. , year 2020 . title Why normalizing flows fail to detect out-of-distribution data . journal NeurIPS volume 33 , pages 20578--20589

    2020

  25. [33]

    , author Lee, H

    author Lee, K. , author Lee, H. , author Lee, K. , author Shin, J. , year 2018 a. title Training confidence-calibrated classifiers for detecting out-of-distribution samples , in: booktitle ICLR

    2018

  26. [34]

    , author Lee, K

    author Lee, K. , author Lee, K. , author Lee, H. , author Shin, J. , year 2018 b. title A simple unified framework for detecting out-of-distribution samples and adversarial attacks . journal NeurIPS volume 31

    2018

  27. [35]

    , author Lee, K

    author Lee, K. , author Lee, K. , author Min, K. , author Zhang, Y. , author Shin, J. , author Lee, H. , year 2018 c. title Hierarchical novelty detection for visual object recognition , in: booktitle CVPR , pp. pages 1034--1042

    2018

  28. [36]

    , author Zhang, X

    author Li, J. , author Zhang, X. , author Yue, K. , author Chen, J. , author Chen, Z. , author Li, W. , year 2024 . title An auto-regulated universal domain adaptation network for uncertain diagnostic scenarios of rotating machinery . journal Expert Systems with Applications v...

    2024

  29. [37]

    , author Li, J

    author Li, X. , author Li, J. , author Zhu, L. , author Wang, G. , author Huang, Z. , year 2021 . title Imbalanced source-free domain adaptation , in: booktitle Proceedings of the 29th ACM International Conference on Multimedia , pp. pages 3330--3339

    2021

  30. [38]

    , author Vasconcelos, N

    author Li, Y. , author Vasconcelos, N. , year 2020 . title Background data resampling for outlier-aware classification , in: booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. pages 13218--13227

    2020

  31. [39]

    , author Li, Y

    author Liang, S. , author Li, Y. , author Srikant, R. , year 2018 . title Enhancing the reliability of out-of-distribution image detection in neural networks , in: booktitle ICLR

    2018

  32. [40]

    , author Cao, Z

    author Liu, H. , author Cao, Z. , author Long, M. , author Wang, J. , author Yang, Q. , year 2019 . title Separate to adapt: Open set domain adaptation via progressive separation , in: booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition ...

    2019

  33. [41]

    , author Hospedales, T

    author Liu, L. , author Hospedales, T. , author LeCun, Y. , author Long, M. , author Luo, J. , author Ouyang, W. , author Pietik \"a inen, M. , author Tuytelaars, T. , year 2024 . title Learning with fewer labels in computer vision . journal IEEE Transactions on Pattern Analys...

    2024

  34. [42]

    , author Wang, X

    author Liu, W. , author Wang, X. , author Owens, J. , author Li, Y. , year 2020 . title Energy-based out-of-distribution detection . journal NeurIPS volume 33 , pages 21464--21475

    2020

  35. [43]

    , author Cao, Y

    author Long, M. , author Cao, Y. , author Wang, J. , author Jordan, M. , year 2015 . title Learning transferable features with deep adaptation networks , in: booktitle International conference on machine learning , organization PMLR . pp. pages 97--105

    2015

  36. [44]

    , author Zhu, H

    author Long, M. , author Zhu, H. , author Wang, J. , author Jordan, M.I. , year 2016 . title Unsupervised domain adaptation with residual transfer networks . journal Advances in neural information processing systems volume 29

    2016

  37. [45]

    , author Zhu, H

    author Long, M. , author Zhu, H. , author Wang, J. , author Jordan, M.I. , year 2017 . title Deep transfer learning with joint adaptation networks , in: booktitle International conference on machine learning , organization PMLR . pp. pages 2208--2217

    2017

  38. [46]

    , author Zhu, K

    author Lu, F. , author Zhu, K. , author Zhai, W. , author Zheng, K. , author Cao, Y. , year 2023 . title Uncertainty-aware optimal transport for semantically coherent out-of-distribution detection . journal arXiv preprint arXiv:2303.10449

    2023 arXiv

  39. [47]

    , author Guo, X

    author Luo, Y. , author Guo, X. , author Liu, L. , author Yuan, Y. , year 2024 . title Dynamic attribute-guided few-shot open-set network for medical image diagnosis . journal Expert Systems with Applications volume 251 , pages 124098

    2024

  40. [48]

    , author Hinton, G

    author Van der Maaten, L. , author Hinton, G. , year 2008 . title Visualizing data using t-sne. journal Journal of machine learning research volume 9

    2008

  41. [49]

    , author Dama s evi c ius, R

    author Maqsood, S. , author Dama s evi c ius, R. , author Shahid, S. , author Forkert, N.D. , year 2024 . title Mox-net: Multi-stage deep hybrid feature fusion and selection framework for monkeypox classification . journal Expert Systems with Applications volume 255 , pages 124584

    2024

  42. [50]

    , author Manrai, A.K

    author Melas-Kyriazi, L. , author Manrai, A.K. , year 2021 . title Pixmatch: Unsupervised domain adaptation via pixelwise consistency training , in: booktitle Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. pages 12435--12445

    2021

  43. [51]

    , author Sun, Y

    author Ming, Y. , author Sun, Y. , author Dia, O. , author Li, Y. , year 2023 . title How to exploit hyperspherical embeddings for out-of-distribution detection? , in: booktitle The Eleventh ICLR

    2023

  44. [52]

    , author Pitale, M

    author Mohseni, S. , author Pitale, M. , author Yadawa, J. , author Wang, Z. , year 2020 . title Self-supervised learning for generalizable out-of-distribution detection , in: booktitle AAAI , pp. pages 5216--5223

    2020

  45. [53]

    , author Damen, D

    author Munro, J. , author Damen, D. , year 2020 . title Multi-modal domain adaptation for fine-grained action recognition , in: booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. pages 122--132

    2020

  46. [54]

    , author Olson, M

    author Neal, L. , author Olson, M. , author Fern, X. , author Wong, W.K. , author Li, F. , year 2018 . title Open set learning with counterfactual images , in: booktitle Proceedings of the European Conference on Computer Vision (ECCV) , pp. pages 613--628

    2018

  47. [55]

    , author Le, T

    author Nguyen, T. , author Le, T. , author Zhao, H. , author Tran, Q.H. , author Nguyen, T. , author Phung, D. , year 2021 . title Most: Multi-source domain adaptation via optimal transport for student-teacher learning , in: booktitle Uncertainty in Artificial Intelligence , o...

    2021

  48. [56]

    , author Nguyen, V

    author Nguyen, T. , author Nguyen, V. , author Le, T. , author Zhao, H. , author Tran, Q.H. , author Phung, D. , year 2022 . title Cycle class consistency with distributional optimal transport and knowledge distillation for unsupervised domain adaptation , in: booktitle Uncert...

    2022

  49. [57]

    , author Gross, S

    author Paszke, A. , author Gross, S. , author Massa, F. , author Lerer, A. , author Bradbury, J. , author Chanan, G. , author Killeen, T. , author Lin, Z. , author Gimelshein, N. , author Antiga, L. , et al., year 2019 . title Pytorch: An imperative style, high-performance dee...

    2019

  50. [58]

    , author Bai, Q

    author Peng, X. , author Bai, Q. , author Xia, X. , author Huang, Z. , author Saenko, K. , author Wang, B. , year 2019 . title Moment matching for multi-source domain adaptation , in: booktitle Proceedings of the IEEE/CVF international conference on computer vision , pp. pages...

    2019

  51. [59]

    , author Li, Y

    author Peng, X. , author Li, Y. , author Saenko, K. , year 2020 . title Domain2vec: Domain embedding for unsupervised domain adaptation , in: booktitle Computer Vision--ECCV 2020: 16th European Conference, Glasgow, UK, August 23--28, 2020, Proceedings, Part VI 16 , organizatio...

    2020

  52. [60]

    , author Usman, B

    author Peng, X. , author Usman, B. , author Kaushik, N. , author Wang, D. , author Hoffman, J. , author Saenko, K. , year 2018 . title Visda: A synthetic-to-real benchmark for visual domain adaptation , in: booktitle Proceedings of the IEEE Conference on Computer Vision and Pa...

    2018

  53. [61]

    , author Khare, S

    author Prabhu, V. , author Khare, S. , author Kartik, D. , author Hoffman, J. , year 2021 . title Sentry: Selective entropy optimization via committee consistency for unsupervised domain adaptation , in: booktitle Proceedings of the IEEE/CVF International Conference on Compute...

    2021

  54. [62]

    , author Ordonez, V

    author Rastegari, M. , author Ordonez, V. , author Redmon, J. , author Farhadi, A. , year 2016 . title Xnor-net: Imagenet classification using binary convolutional neural networks , in: booktitle European conference on computer vision , organization Springer . pp. pages 525--542

    2016

  55. [63]

    , author Runge, J

    author Reimers, C. , author Runge, J. , author Denzler, J. , year 2020 . title Determining the relevance of features for deep neural networks , in: booktitle European Conference on Computer Vision , organization Springer . pp. pages 330--346

    2020

  56. [64]

    , author Zhai, Y

    author Ren, C.X. , author Zhai, Y. , author Luo, Y.W. , author Yan, H. , year 2024 . title Towards unsupervised domain adaptation via domain-transformer . journal International Journal of Computer Vision , pages 1--21

    2024

  57. [65]

    , author Liu, P.J

    author Ren, J. , author Liu, P.J. , author Fertig, E. , author Snoek, J. , author Poplin, R. , author Depristo, M. , author Dillon, J. , author Lakshminarayanan, B. , year 2019 . title Likelihood ratios for out-of-distribution detection . journal NeurIPS volume 32

    2019

  58. [66]

    , author Tian, J

    author Ru, J. , author Tian, J. , author Xiao, C. , author Li, J. , author Shen, H.T. , year 2023 . title Imbalanced open set domain adaptation via moving-threshold estimation and gradual alignment . journal IEEE Transactions on Multimedia

    2023

  59. [67]

    , author Kim, D

    author Saito, K. , author Kim, D. , author Sclaroff, S. , author Saenko, K. , year 2020 . title Universal domain adaptation through self supervision . journal Advances in neural information processing systems volume 33 , pages 16282--16292

    2020

  60. [68]

    , author Saenko, K

    author Saito, K. , author Saenko, K. , year 2021 a. title Ovanet: One-vs-all network for universal domain adaptation , in: booktitle Proceedings of the ieee/cvf international conference on computer vision , pp. pages 9000--9009

    2021

  61. [69]

    , author Saenko, K

    author Saito, K. , author Saenko, K. , year 2021 b. title Ovanet: One-vs-all network for universal domain adaptation , in: booktitle Proceedings of the IEEE/CVF International Conference on Computer Vision , pp. pages 9000--9009

    2021

  62. [70]

    , author Yamamoto, S

    author Saito, K. , author Yamamoto, S. , author Ushiku, Y. , author Harada, T. , year 2018 . title Open set domain adaptation by backpropagation , in: booktitle Proceedings of the European Conference on Computer Vision (ECCV) , pp. pages 153--168

    2018

  63. [71]

    , author \'A lvarez, D

    author Serr \`a , J. , author \'A lvarez, D. , author G \'o mez, V. , author Slizovskaia, O. , author N \'u \ n ez, J.F. , author Luque, J. , year 2019 . title Input complexity and out-of-distribution detection with likelihood-based generative models , in: booktitle ICLR

    2019

  64. [72]

    , author Atwell, K

    author Sicilia, A. , author Atwell, K. , author Alikhani, M. , author Hwang, S.J. , year 2022 . title Pac-bayesian domain adaptation bounds for multiclass learners , in: booktitle Uncertainty in Artificial Intelligence , organization PMLR . pp. pages 1824--1834

    2022

  65. [73]

    , author Zheng, H

    author Sun, J. , author Zheng, H. , author Diao, W. , author Sun, Z. , author Qi, Z. , author Wang, X. , year 2025 . title Prototype-optimized unsupervised domain adaptation via dynamic transformer encoder for sensor drift compensation in electronic nose systems . journal Expe...

    2025

  66. [74]

    , author Guo, C

    author Sun, Y. , author Guo, C. , author Li, Y. , year 2021 . title React: Out-of-distribution detection with rectified activations . journal NeurIPS volume 34 , pages 144--157

    2021

  67. [75]

    , author Li, Y

    author Sun, Y. , author Li, Y. , year 2022 . title Dice: Leveraging sparsification for out-of-distribution detection , in: booktitle European Conference on Computer Vision , organization Springer . pp. pages 691--708

    2022

  68. [76]

    , author Wang, W

    author Sun, Y.X. , author Wang, W. , year 2022 . title Exploiting mixed unlabeled data for detecting samples of seen and unseen out-of-distribution classes , in: booktitle Proceedings of the AAAI Conference on Artificial Intelligence , pp. pages 8386--8394

    2022

  69. [77]

    , author Peng, X

    author Tan, S. , author Peng, X. , author Saenko, K. , year 2020 a. title Class-imbalanced domain adaptation: an empirical odyssey , in: booktitle Computer Vision--ECCV 2020 Workshops: Glasgow, UK, August 23--28, 2020, Proceedings, Part I 16 , organization Springer . pp. pages...

    2020

  70. [78]

    , author Peng, X

    author Tan, S. , author Peng, X. , author Saenko, K. , year 2020 b. title Class-imbalanced domain adaptation: An empirical odyssey , in: booktitle European Conference on Computer Vision , organization Springer . pp. pages 585--602

    2020

  71. [79]

    , author Fan, X

    author Tanwisuth, K. , author Fan, X. , author Zheng, H. , author Zhang, S. , author Zhang, H. , author Chen, B. , author Zhou, M. , year 2021 . title A prototype-oriented framework for unsupervised domain adaptation . journal Advances in Neural Information Processing Systems ...

    2021

  72. [80]

    , author Suganuma, M

    author Techapanurak, E. , author Suganuma, M. , author Okatani, T. , year 2020 . title Hyperparameter-free out-of-distribution detection using cosine similarity , in: booktitle ACCV

    2020

  73. [81]

    , author Flamary, R

    author Turrisi, R. , author Flamary, R. , author Rakotomamonjy, A. , author Pontil, M. , year 2022 . title Multi-source domain adaptation via weighted joint distributions optimal transport , in: booktitle Uncertainty in Artificial Intelligence , organization PMLR . pp. pages 1...

    2022

  74. [82]

    , author Hoffman, J

    author Tzeng, E. , author Hoffman, J. , author Zhang, N. , author Saenko, K. , author Darrell, T. , year 2014 . title Deep domain confusion: Maximizing for domain invariance . journal arXiv preprint arXiv:1412.3474

    2014 arXiv

  75. [83]

    , author An, S

    author Ullah, I. , author An, S. , author Kang, M. , author Chikontwe, P. , author Lee, H. , author Choi, J. , author Park, S.H. , year 2024 . title Video domain adaptation for semantic segmentation using perceptual consistency matching . journal Neural Networks volume 179 , p...

    2024

  76. [84]

    , author Lisboa, P.J

    author Vellido, A. , author Lisboa, P.J. , author Vaughan, J. , year 1999 . title Neural networks in business: a survey of applications (1992--1998) . journal Expert Systems with applications volume 17 , pages 51--70

    1999

  77. [85]

    , author Eusebio, J

    author Venkateswara, H. , author Eusebio, J. , author Chakraborty, S. , author Panchanathan, S. , year 2017 a. title Deep hashing network for unsupervised domain adaptation , in: booktitle Proceedings of the IEEE conference on computer vision and pattern recognition , pp. page...

    2017

  78. [86]

    , author Eusebio, J

    author Venkateswara, H. , author Eusebio, J. , author Chakraborty, S. , author Panchanathan, S. , year 2017 b. title Deep hashing network for unsupervised domain adaptation , in: booktitle Proceedings of the IEEE conference on computer vision and pattern recognition , pp. page...

    2017

  79. [87]

    , author Gaurav, A

    author Vernekar, S. , author Gaurav, A. , author Abdelzad, V. , author Denouden, T. , author Salay, R. , author Czarnecki, K. , year 2019 . title Out-of-distribution detection in classifiers via generation . journal arXiv preprint arXiv:1910.04241

    2019 arXiv

  80. [88]

    , author Jammalamadaka, N

    author Vyas, A. , author Jammalamadaka, N. , author Zhu, X. , author Das, D. , author Kaul, B. , author Willke, T.L. , year 2018 . title Out-of-distribution detection using an ensemble of self supervised leave-out classifiers , in: booktitle ECCV , pp. pages 550--564

    2018

  81. [89]

    , author Li, L

    author Wang, Z. , author Li, L. , author Zeng, C. , author Dong, S. , author Sun, J. , year 2025 . title Slbdetection-net: Towards closed-set and open-set student learning behavior detection in smart classroom of k-12 education . journal Expert Systems with Applications volume...

    2025

  82. [90]

    , author Li, S

    author Xie, M. , author Li, S. , author Gong, K. , author Wang, Y. , author Huang, G. , year 2024 . title Adapting across domains via target-oriented transferable semantic augmentation under prototype constraint . journal International Journal of Computer Vision volume 132 , p...

    2024

  83. [91]

    , author Mahmod, R

    author Yahia, M. , author Mahmod, R. , author Sulaiman, N. , author Ahmad, F. , year 2000 . title Rough neural expert systems . journal Expert Systems with Applications volume 18 , pages 87--99

    2000

  84. [92]

    , author Gao, R

    author Yang, Y. , author Gao, R. , author Xu, Q. , year 2022 a. title Out-of-distribution detection with semantic mismatch under masking , in: booktitle ECCV , organization Springer . pp. pages 373--390

    2022

  85. [93]

    , author Wang, H

    author Yang, Y. , author Wang, H. , author Katabi, D. , year 2022 b. title On multi-domain long-tailed recognition, imbalanced domain generalization and beyond , in: booktitle European Conference on Computer Vision , organization Springer . pp. pages 57--75

    2022

  86. [94]

    , author Yang, X

    author Yang, Y. , author Yang, X. , author Sakamoto, T. , author Fioranelli, F. , author Li, B. , author Lang, Y. , year 2022 c. title Unsupervised domain adaptation for disguised-gait-based person identification on micro-doppler signatures . journal IEEE Transactions on Circu...

    2022

  87. [95]

    , author Chen, B

    author Yao, Y. , author Chen, B. , author Liu, C. , author Feng, C. , author Gao, X. , author Gu, Y. , year 2024 . title Open-set adversarial domain match for electronic nose drift compensation and unknown gas recognition . journal Expert Systems with Applications volume 250 ,...

    2024

  88. [96]

    , author Long, M

    author You, K. , author Long, M. , author Cao, Z. , author Wang, J. , author Jordan, M.I. , year 2019 . title Universal domain adaptation , in: booktitle Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pp. pages 2720--2729

    2019

  89. [97]

    , author Aizawa, K

    author Yu, Q. , author Aizawa, K. , year 2019 . title Unsupervised out-of-distribution detection by maximum classifier discrepancy , in: booktitle ICCV , pp. pages 9518--9526

    2019

  90. [98]

    , author Bisagno, N

    author Zaeemzadeh, A. , author Bisagno, N. , author Sambugaro, Z. , author Conci, N. , author Rahnavard, N. , author Shah, M. , year 2021 . title Out-of-distribution detection using union of 1-dimensional subspaces , in: booktitle CVPR , pp. pages 9452--9461

    2021

  91. [99]

    , author Sch \"o rghuber, M

    author Zendel, O. , author Sch \"o rghuber, M. , author Rainer, B. , author Murschitz, M. , author Beleznai, C. , year 2022 . title Unifying panoptic segmentation for autonomous driving , in: booktitle CVPR , pp. pages 21351--21360

    2022

  92. [100]

    , author Zhang, W

    author Zhang, R. , author Zhang, W. , author Fang, R. , author Gao, P. , author Li, K. , author Dai, J. , author Qiao, Y. , author Li, H. , year 2022 . title Tip-adapter: Training-free adaption of clip for few-shot classification , in: booktitle European Conference on Computer...

    2022

  93. [101]

    , author Lin, J

    author Zhang, T. , author Lin, J. , author Jiao, J. , author Li, H. , year 2024 . title Cross-domain data fusion generation: A novel composite label-guided generative solution for adaptation diagnosis . journal Knowledge-Based Systems volume 301 , pages 112284

    2024

  94. [102]

    , author Deng, B

    author Zhang, Y. , author Deng, B. , author Tang, H. , author Zhang, L. , author Jia, K. , year 2020 . title Unsupervised multi-class domain adaptation: Theory, algorithms, and practice . journal IEEE Transactions on Pattern Analysis and Machine Intelligence volume 44 , pages ...

    2020

  95. [103]

    , year 2022

    author Zhou, Y. , year 2022 . title Rethinking reconstruction autoencoder-based out-of-distribution detection , in: booktitle CVPR , pp. pages 7379--7387

    2022

  96. [104]

    , author Guo, L.Z

    author Zhou, Z. , author Guo, L.Z. , author Cheng, Z. , author Li, Y.F. , author Pu, S. , year 2021 . title Step: Out-of-distribution detection in the presence of limited in-distribution labeled data . journal NeurIPS volume 34 , pages 29168--29180

    2021

Pith tools

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