Pith. sign in

REVIEW 4 major objections 5 minor 51 references

Multi-Granularity Class Prototype Topology Distillation for Class-Incremental Source-Free Unsupervised Domain Adaptation

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read GROTO claims a single algorithm handles class-incremental source-free domain adaptation by mining present classes and distilling source topology, reporting state-of-the-art accuracy on three benchmarks.

desk verdict A genuine new problem setup and a plausible method, but the SOTA claim needs variance estimates and a defined p(f_i) before it convinces. read the letter →

arxiv 2411.16064 v4 pith:TCOYPLH6 submitted 2024-11-25 cs.CV

Peihua Deng , Jiehua Zhang , Xichun Sheng , Chenggang Yan , Yaoqi Sun , Ying Fu , Liang Li This is my paper · ORCID
classification cs.CV
keywords class-incrementallearningsource-freeunsuperviseddomainadaptationprototypetopologydistillationpseudo-labelingcatastrophicforgettingcontrastive
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 tackles class-incremental source-free unsupervised domain adaptation (CI-SFUDA): a model trained on labeled source data arrives at a target domain where unlabeled images of new class subsets appear session by session, with no access to source images or previously seen target images. Its claim is that a single algorithm, GROTO, can keep learning these increments without collapsing old knowledge, by first deciding which source classes are actually present in each session, then pulling target features of the same class together, and finally distilling the source feature topology into the target classifier. This matters because it would let a deployed model be updated on new categories in a privacy-preserving, streaming way instead of being retrained from scratch. The paper reports state-of-the-art final and per-session accuracy on Office-31-CI, Office-Home-CI, and ImageNet-Caltech-CI, with ablations attributing gains to each module.

What carries the argument

Positive classes are first mined by modeling the per-class accumulation distributions of source similarity and target prediction probability, keeping only classes whose values exceed the mean in both signals. The load-bearing objects are multi-granularity class prototypes and the topological structures formed from them. Coarse-grained prototypes are the source classifier weight vectors plus target features closer than the mean cosine-distance threshold to their class centroid, while fine-grained prototypes are target features whose average prediction confidence is above the batch threshold and whose augmented-view confidence has low standard deviation; each unlabeled sample is pseudo-labeled by the positive class with the smallest average cosine distance to its prototypes. The topology is the complete set of pairwise cosine distances between the $N$ source classifier weights $\mu$ and the $N$ target classifier weights $f$. The prototype topology distillation loss $L_{\mathrm{ptd}} = L_{\mathrm{com}} + L_{\mathrm{sep}}$ weights each point-to-point cosine distance by a soft assignment, with target class proportions $p(f_i)$, so that every target prototype is drawn toward the source prototype it matches while source prototypes are not collapsed onto a single target prototype. The self-organization objective $L_{\mathrm{ptfs}} = L_{\mathrm{ce}} + L_{\mathrm{con}}$ uses these pseudo-labels and supervised contrastive learning to make same-class target features cluster and different-class features separate.

What would settle it

Build a synthetic target session with ten classes where one class has only a few samples, such as five percent of the batch, while another absent source class receives high confidence from the frozen source model; run the positive-class mining and measure its precision and recall against the ground-truth present classes, and if the rare class is missed or the absent class is flagged as positive, the above-mean accumulation rule has failed and the downstream pseudo-labels, prototypes, and topology distillation inherit the error.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that the two difficulties of CI-SFUDA can be attacked at the prototype level rather than at the image level. Positive classes, meaning the source classes present in a given target session, are mined by combining two accumulation statistics: the per-class average softmax-normalized similarity between target features and stored source centroids, and the per-class cumulative prediction probability of the frozen source model; a class is positive only if both exceed the mean over all source classes. Reliable pseudo-labels are then generated from multi-granularity prototypes, coarse-grained ones being source classifier weights plus confident target features near class centroids, and fine-grained ones being augmented-consistent confident features, and a cross-entropy plus supervised-contrastive loss organizes the target features. To prevent forgetting, the source and target classifier-weight vectors of the positive classes are treated as point sets, and two distillation losses, a compactness loss and a separability loss, pull each target prototype toward its corresponding source prototype while keeping the target prototypes separable from one another; a small exemplar memory provides replay. With a ViT-B backbone, this yields average final accuracies of 93.0 on Office-31-CI, 82.5 on Office-Home-CI, and 88.8 on ImageNet-Caltech-CI, which the paper reports as state-of-the-art.

Load-bearing premise

The method assumes that, in every target session, the classes actually present can be identified simply as those whose average similarity to source centroids and cumulative target prediction probability both exceed the mean across all source classes; if that fails for a class, its data receives no reliable pseudo-labels and the whole adaptation chain suffers.

Editorial extensions

If this is right

  • A model can be adapted to new target classes in sequence without labeled source data and without replaying old target images, keeping only a small exemplar memory of ten examples per class.
  • Old-class accuracy remains relatively stable across sessions because the classifier is anchored to the source topology instead of drifting toward the latest batch.
  • The method outperforms both class-incremental domain adaptation methods that need source data and source-free universal methods that need all target data available at once.
  • The results across Office-31-CI, Office-Home-CI, ImageNet-Caltech-CI, and the supplementary DomainNet-126-CI suggest the approach scales to larger label spaces.
  • Ablations show that removing the positive-class mining, the self-organization, the topology distillation, or the contrastive loss each reduces final accuracy, indicating every module contributes to the reported gains.

Reading between the lines

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

  • Beyond the paper, the above-mean mining rule is the natural failure point to stress-test: under a heavily imbalanced target session or a poorly calibrated source model, a rare present class could fall below the mean and be dropped, while an absent high-confidence class could be admitted.
  • Beyond the paper, the same two-signal mining and prototype topology distillation could be extended to open-set or novel-class arrivals, where the target label space is not a subset of the source label space, by replacing the mean threshold with a learned rejection boundary.
  • Beyond the paper, because the topology distillation acts only on the $N \times N$ prototype distances, the mechanism could be attached to large foundation-model backbones with lightweight classifier heads without storing source images.
  • Beyond the paper, one could test cutting the contrastive term in later sessions; the ablations show it helps the full pipeline, but removing it after the first session might reduce the reported training overhead.
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

4 major / 5 minor

Summary. The paper proposes GROTO, a method for Class-Incremental Source-Free Unsupervised Domain Adaptation (CI-SFUDA). The target data arrive in sequential sessions with disjoint class subsets of the source label space, and no source data are available during adaptation. GROTO consists of two main modules: (1) multi-granularity class prototype self-organization, which mines positive classes via source-similarity and target-probability accumulation distributions, constructs coarse- and fine-grained prototypes, and trains with cross-entropy and contrastive losses; and (2) prototype topology distillation, which aligns target classifier prototypes to source prototypes through compactness and separability losses to mitigate forgetting. The paper reports state-of-the-art final and session accuracies on Office-31-CI, Office-Home-CI, ImageNet-Caltech-CI, and adds DomainNet-126-CI results in the supplement. Code is released.

Significance. If the reported results are reproducible, GROTO would be a strong baseline for CI-SFUDA, a setting that combines source-free adaptation, class-incremental shifts, and unlabeled target streams. The paper contributes a new problem instantiation and a modular pipeline, with extensive benchmark comparisons, ablations, mining-accuracy tables, and complexity analyses. The release of code is a concrete practical asset. The significance is currently tempered by missing implementation details in the topology distillation loss and by the lack of statistical evidence for the claimed state-of-the-art margins.

major comments (4)
  1. [Sec. 3.4, Eqs. (14)-(15)] The target proportion p(f_i) is never defined. The text states only that p(·) is the target proportion, and Algorithm 4 lists it as an input, but no estimation procedure, no source (e.g., normalized cumulative probabilities from Eq. (2)), and no uniformity assumption are provided. Because p(f_i) weights every term in L_com and L_sep, the loss is not implementable as written. Please specify p(·) explicitly or, if the method is intended to be uniform, state that directly.
  2. [Sec. 3.4 and Algorithm 4] The construction of the source and target prototypes used in L_ptd is incomplete. The source classifier weights are introduced as μ=[μ_1,...,μ_N] for N positive classes, but the paper does not state how the N positive-class columns are selected from the K-class source classifier, nor how the target classifier weights f=[f_1,...,f_N] are initialized at session 1 and expanded when new sessions add classes. This matters because L_ptd compares f against μ; without this specification, the update of the target classifier head is ambiguous.
  3. [Sec. 4.2 and Table 5] The state-of-the-art claim rests on single-run comparisons with small margins (e.g., +1.0 percentage points on ImageNet-Caltech-CI and +1.3 on Office-Home-CI average), and the default hyperparameters in Table 5 (μ0_c=0.5, β=1e-4, nr=10) are selected using the same ImageNet-Caltech-CI average final accuracy that is reported in Table 1. No error bars or multiple seeds are reported. Please report mean±std over at least three seeds for the main comparisons, and either select hyperparameters on a validation split or state clearly that the reported numbers themselves involve test-set selection.
  4. [Sec. 3.3, Eqs. (1)-(2), and Supplementary Tables 7-9] The HKPCM threshold compares each class's source-similarity and cumulative probability to the mean over all K source classes. The supplementary tables show that false positives are not negligible (e.g., Table 7, I→C sessions 3, 4, 6, 7, and 8 have GROTO TCD Acc. of 90.9, 75.0, 64.3, 66.7, and 71.4 versus ProCA-B 100.0, 83.3, 69.2, 83.3, and 81.8). Since every downstream component consumes the mined positive classes, the paper should analyze when the mean-threshold assumption can fail (e.g., imbalanced target batches, overconfident source models, or classes with very different priors) and, at minimum, report the sensitivity of final accuracy to mining errors.
minor comments (5)
  1. [Sec. 3.3, Eq. (2)] The notation P_k = Σ_i C_s(G_s(x_i)) is ambiguous: C_s(G_s(x_i)) is a vector, and the equation should refer to the class-k component C_s^k(G_s(x_i)).
  2. [Eq. (11) and Algorithm 3] There is an inconsistency in the contrastive loss denominator: the main text uses exp(ϕ(G_t(x_i), G_t(x_b))/κ), while Supplementary Algorithm 3 line 43 writes exp(ϕ(G_t(x_i), G_t(x'_i))/κ). Please unify the formula.
  3. [Sec. 4.1] Office-31-CI is described as having 31 shared classes, but the described split covers 30 classes (three disjoint subsets of 10). Clarify whether one source class is intentionally held out as a permanent negative class, and similarly clarify the number of unused source classes in Office-Home-CI and ImageNet-Caltech-CI.
  4. [Sec. 3.3] The sentence 'With the pseudo label assigned in subsequent stages, unique pseudo-labels evolve as the positive classes' is unclear and should be rewritten to state how pseudo-label sets are constructed and expanded across sessions.
  5. [References and figure captions] Reference [35] is incomplete (missing venue and full title), and the captions of Figures 3 and 4 should define the plotted quantities, particularly 'average accuracy of 0 to 9 classes' and the histogram values shown in Figure 4.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the SOTA claim is an empirical result on held-out target test sets, and no component of GROTO reduces by construction to its own inputs.

full rationale

The paper's claimed derivation chain is algorithmic: HKPCM (Sec. 3.3) identifies positive classes from source-similarity and target-probability accumulation statistics (Eqs. 1-2), PTFS generates pseudo-labels via coarse/fine prototypes (Eqs. 3-8), and PTD aligns source and target prototypes (Eqs. 14-16). None of these quantities is defined in terms of the final reported accuracy, and the final accuracy is measured on held-out target test splits from Office-31-CI, Office-Home-CI, and ImageNet-Caltech-CI. The pseudo-labeling loop is standard self-training/bootstrap, not a by-construction reduction: the model's own outputs are used as soft targets, but the reported metric is computed on unseen test data. Comparisons in Tables 1-3 are against external baselines, and the self-citations in the related work (e.g., [2], [45]) are not load-bearing for the method or its evaluation; no uniqueness theorem or ansatz is imported from the authors' prior work. The only evaluation-protocol caveat is test-set hyperparameter selection in Table 5 (e.g., 'the final accuracy is highest when the number of incremental classes γ is 13, but ... we choose 10'), which can inflate the reported number but does not make the derivation equivalent to its inputs; the final accuracy is not a fit parameter. The supplementary material also states a PTD trade-off ('this trade-off may induce the extractor to extract target features ... instead of the most discriminative features of the target domain itself'), which is an acknowledged limitation rather than a circular step. Therefore no circularity is found.

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

The method relies on several domain assumptions about the consistency between source and target feature spaces, plus an undocumented target proportion term in the distillation loss. No new physical entities are introduced; the prototypes and topologies are algorithmic constructs.

free parameters (5)
  • gamma (number of incremental classes per session) = 10 (13 gives highest accuracy on ImageNet-Caltech-CI in Table 5)
    Tuned on the test benchmark; authors chose 10 to match ProCA's setup.
  • mu_0 (initial contrastive loss coefficient) = 0.5
    Tuned in Table 5 on ImageNet-Caltech-CI.
  • beta (exponential decay rate for mu_c) = 1e-4
    Tuned in Table 5 on ImageNet-Caltech-CI.
  • nr (memory exemplars per class) = 10
    Tuned in Table 5 on ImageNet-Caltech-CI.
  • temperature kappa in contrastive loss = not reported
    Appears in Eq. 11 but its value is not specified in the paper.
assumptions (4)
  • domain assumption The source classifier weights can serve as stable topological anchors for target class prototypes across sessions.
    Used in Sec. 3.4 to define Lcom and Lsep; assumes the source feature space is sufficiently aligned with the target feature space for point-to-point prototype distillation to preserve old-class decision boundaries.
  • domain assumption Positive classes are detectable by above-average source similarity and above-average cumulative target probability.
    Core rule of HKPCM in Sec. 3.3; assumes the statistical regularities of the benchmark hold generally.
  • domain assumption The stored source feature centroids and source model are available and correctly represent source class distributions.
    Standard in SFUDA; the method relies on these centroids for both positive class mining and topology distillation.
  • ad hoc to paper Target class proportions p(f_i) in Eqs. 14-15 are known or accurately estimated.
    The formula for p(·) is never given; the loss design depends on this quantity.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Granularity Class Prototype Topology Distillation for Class-Incremental Source-Free Unsupervised Domain Adaptation." pith.science (2026). https://pith.science/paper/TCOYPLH6

@misc{pith2026241116064,
  author       = {Pith},
  title        = {Pith review of: Multi-Granularity Class Prototype Topology Distillation for Class-Incremental Source-Free Unsupervised Domain Adaptation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/TCOYPLH6}},
  note         = {Machine review of arXiv:2411.16064}
}
read the original abstract

This paper explores the Class-Incremental Source-Free Unsupervised Domain Adaptation (CI-SFUDA) problem, where the unlabeled target data come incrementally without access to labeled source instances. This problem poses two challenges, the interference of similar source-class knowledge in target-class representation learning and the shocks of new target knowledge to old ones. To address them, we propose the Multi-Granularity Class Prototype Topology Distillation (GROTO) algorithm, which effectively transfers the source knowledge to the class-incremental target domain. Concretely, we design the multi-granularity class prototype self-organization module and the prototype topology distillation module. First, we mine the positive classes by modeling accumulation distributions. Next, we introduce multi-granularity class prototypes to generate reliable pseudo-labels, and exploit them to promote the positive-class target feature self-organization. Second, the positive-class prototypes are leveraged to construct the topological structures of source and target feature spaces. Then, we perform the topology distillation to continually mitigate the shocks of new target knowledge to old ones. Extensive experiments demonstrate that our proposed method achieves state-of-the-art performance on three public datasets. Code is available at https://github.com/dengpeihua/GROTO.

Figures

Figures reproduced from arXiv: 2411.16064 by the authors.

Figure 1
Figure 1. An illustration of Class-Incremental Source-Free Un [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. An overview of the GROTO algorithm, it includes two modules: 1) Multi-granularity class prototype self-organization: we mine [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. The accuracies on old and new classes of GROTO [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (6 more)
Figure 4
Figure 4. Figure 4: The target cumulative probability and source average [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The multi-granularity class prototype visualizations of [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: The feature distribution and decision boundary visual [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Feature distribution and decision boundary visualizations of “GROTO w/o PTD” and “GROTO w/ PTD” models on Office-31-CI. [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: Source and target feature distribution visualizations of “GROTO w/o PTD” and “GROTO w/ PTD” models. [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]
Figure 9
Figure 9. Figure 9: Target feature distribution of source-only (ViT-B), ProCA-B, LEAD-B, and GROTO methods on Office-31-CI. [PITH_FULL_IMAGE:figures/full_fig_p018_9.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

51 extracted references · 43 canonical work pages

  1. [1]

    Learning to transfer examples for partial domain adaptation

    Zhangjie Cao, Kaichao You, Mingsheng Long, Jianmin Wang, and Qiang Yang. Learning to transfer examples for partial domain adaptation. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 2985–2994, 2019. 1

    work page 2019

  2. [2]

    Yiming Cui, Liang Li, Jiehua Zhang, Chenggang Yan, Hongkui Wang, Shuai Wang, Heng Jin, and Li Wu. Stochas- A→W Source-only: ViT-B CIUDA: ProCA-B SF-UniDA: LEAD-B CI-SFUDA: GROTO A→D Source-only: ViT-B CIUDA: ProCA-B SF-UniDA: LEAD-B CI-SFUDA: GROTO D→W Source-only: ViT-B CIUDA: ProCA-B SF-UniDA: LEAD-B CI-SFUDA: GROTO D→A Source-only: ViT-B CIUDA: ProCA-B...

    work page 2024

  3. [3]

    A continual learning survey: Defying for- getting in classification tasks

    Matthias De Lange, Rahaf Aljundi, Marc Masana, Sarah Parisot, Xu Jia, Ale ˇs Leonardis, Gregory Slabaugh, and Tinne Tuytelaars. A continual learning survey: Defying for- getting in classification tasks. IEEE transactions on pattern analysis and machine intelligence , 44(7):3366–3385, 2021. 5

    work page 2021

  4. [4]

    An image is worth 16x16 words: Trans- formers for image recognition at scale

    Alexey Dosovitskiy, Lucas Beyer, Alexander Kolesnikov, Dirk Weissenborn, Xiaohua Zhai, Thomas Unterthiner, Mostafa Dehghani, Matthias Minderer, Georg Heigold, Syl- vain Gelly, et al. An image is worth 16x16 words: Trans- formers for image recognition at scale. arXiv preprint arXiv:2010.11929, 2020. 6, 5

    arXiv 2010

  5. [5]

    Caltech-256 object category dataset, 2007

    Gregory Griffin, Alex Holub, and Pietro Perona. Caltech-256 object category dataset, 2007. CIT Technical Report, 7694,

    work page 2007

  6. [6]

    Gradient reweighting: Towards imbalanced class-incremental learning

    Jiangpeng He. Gradient reweighting: Towards imbalanced class-incremental learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pages 16668–16677, 2024. 2

    work page 2024

  7. [7]

    Class-incremental learning with clip: Adaptive representa- tion adjustment and parameter fusion

    Linlan Huang, Xusheng Cao, Haori Lu, and Xialei Liu. Class-incremental learning with clip: Adaptive representa- tion adjustment and parameter fusion. In European Confer- ence on Computer Vision , pages 214–231. Springer, 2024. 2

    work page 2024

  8. [8]

    Ovor: Oneprompt with virtual outlier regularization for rehearsal-free class-incremental learning

    Wei-Cheng Huang, Chun-Fu Chen, and Hsiang Hsu. Ovor: Oneprompt with virtual outlier regularization for rehearsal-free class-incremental learning. arXiv preprint arXiv:2402.04129, 2024. 2

    arXiv 2024

Show all 51 references
  1. [9]

    Improved cross-corpus speech emotion recognition using deep local domain adaptation

    Zhao Huijuan, Ye Ning, and Wang Ruchuan. Improved cross-corpus speech emotion recognition using deep local domain adaptation. Chinese Journal of Electronics , 32(3): 640–646, 2023. 1

    2023

  2. [10]

    Class- incremental learning by knowledge distillation with adaptive feature consolidation

    Minsoo Kang, Jaeyoo Park, and Bohyung Han. Class- incremental learning by knowledge distillation with adaptive feature consolidation. In Proceedings of the IEEE/CVF con- ference on computer vision and pattern recognition , pages 16071–16080, 2022. 2

    2022

  3. [11]

    C-sfda: A curriculum learning aided self- training framework for efficient source free domain adapta- tion

    Nazmul Karim, Niluthpol Chowdhury Mithun, Abhinav Ra- jvanshi, Han-pang Chiu, Supun Samarasekera, and Nazanin Rahnavard. C-sfda: A curriculum learning aided self- training framework for efficient source free domain adapta- tion. In Proceedings of the IEEE/CVF Conference on Com...

  4. [12]

    Supervised contrastive learning

    Prannay Khosla, Piotr Teterwak, Chen Wang, Aaron Sarna, Yonglong Tian, Phillip Isola, Aaron Maschinot, Ce Liu, and Dilip Krishnan. Supervised contrastive learning. Advances in neural information processing systems , 33:18661–18673,

  5. [13]

    Class- incremental domain adaptation

    Jogendra Nath Kundu, Rahul Mysore Venkatesh, Naveen Venkat, Ambareesh Revanur, and R Venkatesh Babu. Class- incremental domain adaptation. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23– 28, 2020, Proceedings, Part XIII 16 , pages 53–69. Springer,

    2020

  6. [14]

    Effective decision boundary learning for class incremen- tal learning

    Kunchi Li, Jun Wan, Sergio Escalera, Zhen Lei, and Shan Yu. Effective decision boundary learning for class incremen- tal learning. 2023. 5

    2023

  7. [15]

    Principal properties attention matching for par- tial domain adaptation in fault diagnosis

    Shugang Li, Renhu Bu, Shuang Li, Chi Harold Liu, and Keke Huang. Principal properties attention matching for par- tial domain adaptation in fault diagnosis. IEEE Transactions on Instrumentation and Measurement, 73:1–12, 2024. 1

    2024

  8. [16]

    Prototype- guided continual adaptation for class-incremental unsuper- vised domain adaptation

    Hongbin Lin, Yifan Zhang, Zhen Qiu, Shuaicheng Niu, Chuang Gan, Yanxia Liu, and Mingkui Tan. Prototype- guided continual adaptation for class-incremental unsuper- vised domain adaptation. In European Conference on Com- puter Vision, pages 351–368. Springer, 2022. 1, 2, 5, 6

    2022

  9. [17]

    Class incremental learning via likelihood ratio based task prediction

    Haowei Lin, Yijia Shao, Weinan Qian, Ningxin Pan, Yiduo Guo, and Bing Liu. Class incremental learning via likelihood ratio based task prediction. arXiv preprint arXiv:2309.15048, 2023. 2

    2023 arXiv

  10. [18]

    Guid- ing pseudo-labels with uncertainty estimation for source- free unsupervised domain adaptation

    Mattia Litrico, Alessio Del Bue, and Pietro Morerio. Guid- ing pseudo-labels with uncertainty estimation for source- free unsupervised domain adaptation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 7640–7650, 2023. 6

    2023

  11. [19]

    Exploiting fine-grained pro- totype distribution for boosting unsupervised class incremen- tal learning

    Jiaming Liu, Hongyuan Liu, Zhili Qin, Wei Han, Yulu Fan, Qinli Yang, and Junming Shao. Exploiting fine-grained pro- totype distribution for boosting unsupervised class incremen- tal learning. arXiv preprint arXiv:2408.10046, 2024. 2

    2024 arXiv

  12. [20]

    Entity-enhanced adaptive re- construction network for weakly supervised referring expres- sion grounding

    Xuejing Liu, Liang Li, Shuhui Wang, Zhengjun Zha, Dechao Meng, and Qingming Huang. Entity-enhanced adaptive re- construction network for weakly supervised referring expres- sion grounding. IEEE Transactions on Pattern Analysis and Machine Intelligence, 45:3003–3018, 2022. 1

    2022

  13. [21]

    Model behavior preserving for class-incremental learning

    Yu Liu, Xiaopeng Hong, Xiaoyu Tao, Songlin Dong, Jin- gang Shi, and Yihong Gong. Model behavior preserving for class-incremental learning. IEEE Transactions on Neural Networks and Learning Systems , 34(10):7529–7540, 2022. 2

    2022

  14. [22]

    Source-free domain adap- tation with domain generalized pretraining for face anti- spoofing

    Yuchen Liu, Yabo Chen, Wenrui Dai, Mengran Gou, Chun- Ting Huang, and Hongkai Xiong. Source-free domain adap- tation with domain generalized pretraining for face anti- spoofing. IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, 2024. 2

    2024

  15. [23]

    Diffclass: Diffusion-based class incremental learning

    Zichong Meng, Jie Zhang, Changdi Yang, Zheng Zhan, Pu Zhao, and Yanzhi Wang. Diffclass: Diffusion-based class incremental learning. In European Conference on Computer Vision, pages 142–159. Springer, 2024. 2

    2024

  16. [24]

    Un- derstanding and improving source-free domain adaptation from a theoretical perspective

    Yu Mitsuzumi, Akisato Kimura, and Hisashi Kashima. Un- derstanding and improving source-free domain adaptation from a theoretical perspective. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 28515–28524, 2024. 1

    2024

  17. [25]

    Efficient test-time model adaptation without forgetting

    Shuaicheng Niu, Jiaxiang Wu, Yifan Zhang, Yaofo Chen, Shijian Zheng, Peilin Zhao, and Mingkui Tan. Efficient test-time model adaptation without forgetting. In Interna- tional conference on machine learning, pages 16888–16905. PMLR, 2022. 1

    2022

  18. [26]

    Bmd: A general class-balanced multicen- tric dynamic prototype strategy for source-free domain adap- tation

    Sanqing Qu, Guang Chen, Jing Zhang, Zhijun Li, Wei He, and Dacheng Tao. Bmd: A general class-balanced multicen- tric dynamic prototype strategy for source-free domain adap- tation. In European conference on computer vision , pages 165–182. Springer, 2022. 1, 2

    2022

  19. [27]

    Upcycling models under domain and category shift

    Sanqing Qu, Tianpei Zou, Florian R ¨ohrbein, Cewu Lu, Guang Chen, Dacheng Tao, and Changjun Jiang. Upcycling models under domain and category shift. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 20019–20028, 2023. 2

    2023

  20. [28]

    Lead: Learn- ing decomposition for source-free universal domain adapta- tion

    Sanqing Qu, Tianpei Zou, Lianghua He, Florian R ¨ohrbein, Alois Knoll, Guang Chen, and Changjun Jiang. Lead: Learn- ing decomposition for source-free universal domain adapta- tion. In Proceedings of the IEEE/CVF Conference on Com- puter Vision and Pattern Recognition , pages 2...

  21. [29]

    Fscil-eaca: Few-shot class-incremental learning network based on em- bedding augmentation and classifier adaptation for image classification

    ZHANG Ruru, E Haihong, and SONG Meina. Fscil-eaca: Few-shot class-incremental learning network based on em- bedding augmentation and classifier adaptation for image classification. Chinese Journal of Electronics , 33(1):139– 152, 2024. 2

    2024

  22. [30]

    Imagenet large scale visual recognition challenge

    Olga Russakovsky, Jia Deng, Hao Su, Jonathan Krause, San- jeev Satheesh, Sean Ma, Zhiheng Huang, Andrej Karpathy, Aditya Khosla, Michael Bernstein, et al. Imagenet large scale visual recognition challenge. International journal of computer vision, 115:211–252, 2015. 6

    2015

  23. [31]

    Adapting visual category models to new domains

    Kate Saenko, Brian Kulis, Mario Fritz, and Trevor Darrell. Adapting visual category models to new domains. In Com- puter Vision–ECCV 2010: 11th European Conference on Computer Vision, Heraklion, Crete, Greece, September 5- 11, 2010, Proceedings, Part IV 11, pages 213–226. Springer,

    2010

  24. [32]

    A survey on image data augmentation for deep learning

    Connor Shorten and Taghi M Khoshgoftaar. A survey on image data augmentation for deep learning. Journal of big data, 6(1):1–48, 2019. 5

    2019

  25. [33]

    Source-free domain adaptation via target prediction distribution searching

    Song Tang, An Chang, Fabian Zhang, Xiatian Zhu, Mao Ye, and Changshui Zhang. Source-free domain adaptation via target prediction distribution searching. International jour- nal of computer vision , 132(3):654–672, 2024. 6

    2024

  26. [34]

    Unified source-free domain adaptation

    Song Tang, Wenxin Su, Mao Ye, Jianwei Zhang, and Xiatian Zhu. Unified source-free domain adaptation. arXiv preprint arXiv:2403.07601, 2024. 2, 6, 3, 5

    2024

  27. [35]

    Source- free domain adaptation with frozen multimodal foundation model, 2024

    Song Tang, Wenxin Su, Mao Ye, and Xiatian Zhu. Source- free domain adaptation with frozen multimodal foundation model, 2024. 1, 6, 3, 5

    2024

  28. [36]

    A prototype-oriented framework for unsupervised domain adaptation

    Korawat Tanwisuth, Xinjie Fan, Huangjie Zheng, Shu- jian Zhang, Hao Zhang, Bo Chen, and Mingyuan Zhou. A prototype-oriented framework for unsupervised domain adaptation. Advances in Neural Information Processing Sys- tems, 34:17194–17208, 2021. 5

    2021

  29. [37]

    Topology-preserving class-incremental learning

    Xiaoyu Tao, Xinyuan Chang, Xiaopeng Hong, Xing Wei, and Yihong Gong. Topology-preserving class-incremental learning. In Computer Vision–ECCV 2020: 16th European Conference, Glasgow, UK, August 23–28, 2020, Proceed- ings, Part XIX 16, pages 254–270. Springer, 2020. 2

    2020

  30. [38]

    Transformer-based under- sampled single-pixel imaging

    Ye Tian, Ying Fu, and Jun Zhang. Transformer-based under- sampled single-pixel imaging. Chinese Journal of Electron- ics, 32(5):1151–1159, 2023. 2

    2023

  31. [39]

    SMART: Syntax-Calibrated Multi-Aspect Relation Transformer for Change Captioning

    Yunbin Tu, Liang Li, Li Su, Zheng-Jun Zha, and Qingming Huang. SMART: Syntax-Calibrated Multi-Aspect Relation Transformer for Change Captioning . IEEE Transactions on Pattern Analysis & Machine Intelligence , 46(07):4926– 4943, 2024. 1

    2024

  32. [40]

    Deep hashing network for unsupervised domain adaptation

    Hemanth Venkateswara, Jose Eusebio, Shayok Chakraborty, and Sethuraman Panchanathan. Deep hashing network for unsupervised domain adaptation. In Proceedings of the IEEE conference on computer vision and pattern recogni- tion, pages 5018–5027, 2017. 6

    2017

  33. [41]

    Large scale incre- mental learning

    Yue Wu, Yinpeng Chen, Lijuan Wang, Yuancheng Ye, Zicheng Liu, Yandong Guo, and Yun Fu. Large scale incre- mental learning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , pages 374–382,

  34. [42]

    Unraveling the mysteries of label noise in source-free domain adaptation: Theory and practice

    Gezheng Xu, Li Yi, Pengcheng Xu, Jiaqi Li, Ruizhi Pu, Changjian Shui, Charles Ling, A Ian McLeod, and Boyu Wang. Unraveling the mysteries of label noise in source-free domain adaptation: Theory and practice. IEEE Transactions on Pattern Analysis and Machine Intelligence , 2025. 1

    2025

  35. [43]

    Unsupervised cross-media hashing learning via knowledge graph

    Zhaoda Ye, Xiangteng He, and Yuxin Peng. Unsupervised cross-media hashing learning via knowledge graph. Chinese Journal of Electronics, 31(6):1081–1091, 2022. 2

    2022

  36. [44]

    Inductive state- relabeling adversarial active learning with heuristic clique rescaling

    Beichen Zhang, Liang Li, Shuhui Wang, Shaofei Cai, Zheng- Jun Zha, Qi Tian, and Qingming Huang. Inductive state- relabeling adversarial active learning with heuristic clique rescaling. IEEE Transactions on Pattern Analysis and Ma- chine Intelligence, 46(12):9780–9796, 2024. 2

    2024

  37. [45]

    Monocular depth estimation on adverse weath- ers with curriculum domain distribution alignment

    Jiehua Zhang, Liang Li, Chenggang Yan, Wei Ke, and Yi- hong Gong. Monocular depth estimation on adverse weath- ers with curriculum domain distribution alignment. IEEE Transactions on Circuits and Systems for Video Technology ,

  38. [46]

    Deep guided attention network for joint denoising and demosaicing in real image

    Tao Zhang, Ying Fu, and Jun Zhang. Deep guided attention network for joint denoising and demosaicing in real image. Chinese Journal of Electronics, 33(1):303–312, 2024. 1

    2024

  39. [47]

    From speaker to dubber: movie dubbing with prosody and duration consistency learning

    Zhedong Zhang, Liang Li, Gaoxiang Cong, Haibing Yin, Yuhan Gao, Chenggang Yan, Anton van den Hengel, and Yuankai Qi. From speaker to dubber: movie dubbing with prosody and duration consistency learning. In Proceedings of the 32nd ACM International Conference on Multimedia , pa...

    2024

  40. [48]

    Comparing proba- bility distributions with conditional transport

    Huangjie Zheng and Mingyuan Zhou. Comparing proba- bility distributions with conditional transport. stat, 1050:9,

  41. [49]

    Gait recognition in the wild with dense 3d representations and A benchmark

    Jinkai Zheng, Xinchen Liu, Wu Liu, Lingxiao He, Cheng- gang Yan, and Tao Mei. Gait recognition in the wild with dense 3d representations and A benchmark. In CVPR, pages 20228–20237, 2022. 2

    2022

  42. [50]

    Source-free domain adaptation with class prototype discovery

    Lihua Zhou, Nianxin Li, Mao Ye, Xiatian Zhu, and Song Tang. Source-free domain adaptation with class prototype discovery. Pattern recognition, 145:109974, 2024. 2

    2024

  43. [51]

    Unsupervised domain adaption harnessing vision-language pre-training

    Wenlve Zhou and Zhiheng Zhou. Unsupervised domain adaption harnessing vision-language pre-training. IEEE Transactions on Circuits and Systems for Video Technology ,

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.