Pith. sign in

REVIEW 3 major objections 6 minor 2 cited by

Prompt as Free Lunch: Enhancing Diversity in Source-Free Cross-domain Few-shot Learning through Semantic-Guided Prompting

T0 review · 3 major / 6 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 SeGD-VPT claims that class-description text prompts can act as a free source of diversity, letting a frozen CLIP model reach state-of-the-art source-free cross-domain few-shot accuracy.

desk verdict Solid incremental source-free CD-FSL paper; the SOTA claim is confounded by backbone mismatch, and the semantic-guidance mechanism still needs a random-text control before the 'free lunch' is established. read the letter →

arxiv 2412.00767 v1 pith:FSN24AT4 submitted 2024-12-01 cs.CV cs.CLcs.LG

Linhai Zhuo , Zheng Wang , Yuqian Fu , Tianwen Qian This is my paper · ORCID
classification cs.CVcs.CLcs.LG
keywords source-freecross-domainfew-shotlearningvisualprompttuningdiversitypromptssemanticguidanceCLIPtext-guidedfeaturegenerationtwo-stagetrainingBSCDbenchmark
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper claims that in source-free cross-domain few-shot learning, a frozen CLIP model can be transferred to a new visual domain using only a handful of labeled images by treating text descriptions as a free source of diversity. It proposes SeGD-VPT, which adds learnable diversity prompt tokens to each support image, generates many semantic features from randomly combined class descriptions, and uses those features to guide the prompt learning with a targeted supervised contrastive loss. After this generation phase, the produced prompt visual features train a classifier with ArcFace loss; inference uses only the visual encoder with deep prompts and the classifier. On the BSCD benchmarks, it reports averages of 58.31% and 66.76% in 5-way 1-shot and 5-shot settings, the best among source-free methods and comparable to source-utilized state of the art. If right, it shows text modality can substitute for source-domain data in few-shot transfer.

What carries the argument

The central object is the diversity prompt, a learnable token appended to each support image's token sequence at the input of a frozen CLIP visual transformer. It is paired with deep prompt tokens inserted in each transformer layer, and with diversity semantic features built by randomly combining class-description text features through an adapter. The machinery works by optimizing three losses: a diversity loss that pushes different prompts apart, a semantic contrastive loss that aligns combined text features with class prompts, and a targeted supervised contrastive loss that aligns each prompt visual feature with randomly selected nearest same-class text features. The two-phase schedule separates the diversity-generation objective from the classification objective.

What would settle it

Take a target domain and replace the per-class descriptions with a single generic description shared by all classes, or with descriptions whose class labels are shuffled, then run the SeGD-VPT pipeline unchanged. If accuracy on EuroSAT or CropDisease does not fall well below the reported 83.58% and 90.45% in 1-shot, the claim that semantic description diversity drives the improvement is falsified.

Watch

Extended reading notes

Core claim

The paper's central claim is that sample diversity, not just model capacity, is the bottleneck when adapting a large pretrained vision-language model to a target domain with few samples, and that textual descriptions provide a cheap, domain-consistent way to generate that diversity. SeGD-VPT materializes this by prepending independent learnable diversity prompts to augmented copies of each support image, building diversity semantic features from random combinations of class-description text features aligned to standard class prompts, and guiding the diversity prompts with a targeted supervised contrastive loss that pulls each visual feature toward randomly selected nearest text features of the same class. The resulting prompt visual features are then used to train an ArcFace-based classifier. The paper reports that this beats all source-free CD-FSL baselines and matches or exceeds source-utilized methods, with the largest gains on EuroSAT and CropDisease.

Load-bearing premise

The method's gains depend on the class descriptions harvested from the target domain being semantically accurate and representative; if those descriptions are noisy, generic, or mismatched to the visual content, the semantic guidance loss will drag the diversity prompts toward wrong text landmarks and the generated features will be less useful.

Editorial extensions

If this is right

  • If SeGD-VPT is correct, source-free CD-FSL can be solved by a frozen CLIP backbone plus a handful of learnable tokens, eliminating the need for source-domain data and meta-training.
  • The two-stage design matters: merging the diversity-generation and classification stages into one training drops the 5-shot average from 66.76% to 51.45%, so diversity and discrimination objectives should not be optimized jointly.
  • Generated prompt visual features act as an augmentation that any downstream classifier can consume, which means the first stage is reusable across different classifiers.
  • The largest gains appear in 1-shot and in EuroSAT and CropDisease, suggesting text guidance is most valuable when visual evidence is scarcest and when the domain gap with CLIP pretraining is large.

Reading between the lines

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

  • We infer that SeGD-VPT's upper bound is set by the quality and diversity of the class descriptions; if descriptions are generic or wrong, the semantic guidance would pull diversity prompts toward irrelevant text landmarks, so automatic description generation should be evaluated for robustness.
  • The method suggests a testable extension: replacing language-model-sourced descriptions with descriptions generated on the fly from the target images themselves could make the framework fully self-contained and adaptive to new domains.
  • Because inference discards the text encoder and adapter, the test-time cost is the same as a prompted CLIP classifier; one could compare SeGD-VPT against a version with transductive query-set refinement to see whether the reported gap is due to the inductive setting or to the diversity mechanism.
  • The diversity-loss formulation is a generic regularizer, so it could be applied to other prompt-tuning or adapter methods in few-shot settings to prevent prompt collapse, independent of the semantic guidance.
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

3 major / 6 minor

Summary. The manuscript introduces SeGD-VPT, a source-free cross-domain few-shot learning method built on CLIP. In a first phase it trains per-input diversity prompts and deep prompt tokens, using text descriptions of the target classes to form diversity semantic features and enforcing alignment through the losses L_div, L_se, and L_TSC (Eqs. 1, 3, 4). In a second phase it uses the generated prompt visual features to train an ArcFace classifier. Experiments on ChestX, ISIC, EuroSAT, and CropDisease report 58.31% and 66.76% average accuracy for 5-way 1-shot and 5-way 5-shot settings, and the paper claims the best performance under the source-free CD-FSL setting.

Significance. If the central claim holds, the work makes a practical contribution: it suggests that text-guided prompt diversity can improve CLIP transfer with only a handful of target samples and no source data, which is relevant for medical and satellite-image domains. The paper's positive features include a clear two-step design, careful ablations against CLIP-base and incremental baselines, a feature-count analysis, and a t-SNE visualization supporting the claim of expanded class-conditional distributions. However, the two load-bearing claims—that the gains come from semantic guidance rather than added stochasticity, and that the method is state of the art among source-free methods—are not yet supported by the experimental design as presented.

major comments (3)
  1. [Section 3.3, Table 2] The semantic-guidance mechanism is load-bearing: Equations (2)-(4) train the diversity prompts against diversity semantic features F_s built from ChatGPT-style class descriptions, and Table 2 credits Describe-P with the SeGD-b2 to SeGD-VPT gain of 0.93% average (57.38 to 58.31). The paper does not test whether this gain comes from the semantic content or simply from the additional stochasticity in describe-prompt sampling. I ask for three ablations: (i) use only the class-name prompt '[Domain] photo of [Class]' as the text guidance instead of the full descriptions; (ii) replace the real descriptions with randomly shuffled or generic descriptions of the same length; (iii) remove the top-c/gamma random selection while keeping the same number of text features. Without at least (i), the title-level claim that semantics provides a 'free lunch' is not established. The concern is not circularity; it is that the result may be attributable to a confound.
  2. [Section 4.1, Table 1] The source-free state-of-the-art comparison is confounded by backbone. SeGD-VPT uses a frozen ViT-B/16 CLIP (DFN2B), while the closest source-free competitors VDB and IM-DCL use ResNet-18 and ResNet-10, and the strongest source-utilized ViT baselines use DINO rather than CLIP. Since CLIP is known to transfer strongly to downstream classification, the reported average margins (for example, +2.40% over IM-DCL in 1-shot) may reflect the backbone and pretraining choice rather than the proposed method. To support the claim, report a frozen CLIP ViT-B/16 linear-probe or simple visual-prompt-tuning reference under the same protocol, with the same episodes, query size, and fine-tuning budget.
  3. [Supplementary Table 1; Section 4.1] The per-dataset tuning of training epochs (40, 55, or 60) and learning rates (0.001 versus 0.0001) is reported only in the supplementary material, and no sensitivity study is provided. Because the comparison includes methods that are presumably run under their own original settings, the paper should either fix a single hyperparameter schedule and report the result, or provide a sensitivity table showing that the conclusions are stable. Currently the 'best performance under the source-free CD-FSL setting' claim rests on a favorable per-dataset configuration.
minor comments (6)
  1. [Section 3.2] The text refers to 'Algorithm ??' but no algorithm is present in the main text or the supplementary material; include the pseudo-code or remove the reference.
  2. [Equation (1)] Equation (1) is garbled in the provided PDF (the formula contains repeated '/ba√︂ex' tokens); it should be typeset cleanly so that the cosine-similarity definition is readable.
  3. [Section 4.1] The text says 'totally 14 methods' but then enumerates 16 method names (GNN, FWT, LRP, ATA, ATA-FT, AFA, wave-SAN, StyleAdv, StyleAdv-FT, DARA, Fine-tune, NSAE, BSR, PMF, VDB, IM-DCL); correct the count or the list.
  4. [Table 2] SeGD-b2 is only described informally in the final bullet of Section 4.2; define its exact configuration (diversity prompts without description guidance) before presenting its results.
  5. [Abstract and Section 4.1] There are minor language issues: 'utilize' should be 'utilizes', 'CLP model' should be 'CLIP model' in the abstract, and 'Concretly' should be 'Concretely' in Section 4.1.
  6. [References] References [42] and [43] appear to be the same paper ('Cross-domain few-shot classification via adversarial task augmentation'); if they are different publications, disambiguate them, otherwise consolidate.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity found: SeGD-VPT's reported accuracies come from held-out query evaluation, and the semantic descriptions are external inputs, not fitted outputs.

full rationale

The paper's central claim is an empirical accuracy result on the BSCD benchmarks (Table 1), obtained by training on support samples and evaluating on query samples. The semantic guidance chain in Section 3.3 uses class descriptions ('collected from the target domain, e.g., via ChatGPT') as external inputs to the text encoder, then aligns diversity prompts to these text features; the descriptions are not derived from the query labels or from the reported accuracy, so no equation reduces to its own output. The 'free lunch' claim is an interpretation of the ablation gains, and the absence of a description-quality sensitivity test is an experimental robustness concern, not a circularity reduction. The only self-citations (e.g., TGDM [55], ME-D2N [8], and [3]) appear in related-work positioning and are not load-bearing for the SeGD-VPT design or its results. The method is self-contained against external benchmarks, so the circularity score is 0.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central claim rests on several hand-chosen hyperparameters (sample count 25, deep prompt tokens 5, gamma sampling, per-dataset epochs and learning rates) and on the assumption that text descriptions of target classes are accurate and aligned with CLIP's joint embedding. No new physical or external entities are introduced; the diversity prompts are internal learnable parameters with no independent falsifiable handle outside the paper's experiments.

free parameters (8)
  • Total fine-tuning sample count per task = 25
    Hand-set normalization in Section 4 ('Network Modules'): each support image gets 24 diversity prompts in 1-shot and 4 in 5-shot so that 25 samples are used per task. This determines the amount of generated features and affects accuracy.
  • Deep prompt tokens per layer = 5
    Hand-chosen in Section 4 ('Network Modules'); controls the number of learnable parameters in the frozen CLIP encoder.
  • Diversity prompt length l = 1
    Hand-chosen in Section 4 ('Implementation Details'); length of the learnable prompt vector prepended to image tokens.
  • Top-c and sampled text feature count m = c=300, m=100
    Hand-chosen in Section 4 ('Implementation Details') for selecting and sampling diversity semantic features in the TSC loss.
  • Gamma distribution parameters for random text selection = Gamma(2.0, 75)
    Hand-chosen in Section 4 ('Implementation Details') to introduce randomness when sampling text features.
  • Contrastive temperature tau = 0.07
    Fixed temperature in Equation 4, a standard value but still a hand-chosen constant.
  • Per-dataset training epochs and learning rates = T=40/55/60, lr=0.001/0.0001
    Tuned separately for each of the four datasets (supplementary Table 1), a form of test-set tuning that can inflate reported accuracy.
  • Augmentation count nv, description count ns, combination count ts
    Mentioned in Section 3.3 (N x K x nv variants, ns prompts per class, ts combinations) but their values are never given, making reproduction impossible and allowing hidden tuning.
assumptions (5)
  • domain assumption CLIP's image and text encoders share an aligned embedding space.
    The method relies on cosine similarity between visual and text features in the frozen CLIP space (Sections 3.2, 3.3).
  • domain assumption Target class descriptions gathered via ChatGPT or the web are semantically correct and domain-consistent.
    The semantic guidance loss assumes these descriptions align with the images in each target domain (Section 3.3).
  • ad hoc to paper Minimizing cosine similarity between diversity prompts yields diverse yet useful features rather than degenerate solutions.
    L_div (Equation 1) encourages distinct prompts, but the paper provides no proof or external evidence that this improves classification rather than adding noise; only the empirical ablation supports it.
  • ad hoc to paper The two-step training regime with conflicting objectives is necessary and stable.
    The paper justifies the two-step design by the contradictory goals of diversity (step 1) and consistency (step 2) in Section 4.3; this is a design choice without theoretical guarantee.
  • domain assumption Frozen CLIP backbone pretrained on DFN2B provides useful features for all four target domains.
    The method builds on the CLIP ViT-B/16 pretrained on DFN2B and assumes its features are transferable; the paper notes this fails for ChestX, where accuracy is low (Section 4.1 and Table 1).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Prompt as Free Lunch: Enhancing Diversity in Source-Free Cross-domain Few-shot Learning through Semantic-Guided Prompting." pith.science (2026). https://pith.science/paper/FSN24AT4

@misc{pith2026241200767,
  author       = {Pith},
  title        = {Pith review of: Prompt as Free Lunch: Enhancing Diversity in Source-Free Cross-domain Few-shot Learning through Semantic-Guided Prompting},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FSN24AT4}},
  note         = {Machine review of arXiv:2412.00767}
}
read the original abstract

The source-free cross-domain few-shot learning (CD-FSL) task aims to transfer pretrained models to target domains utilizing minimal samples, eliminating the need for source domain data. Addressing this issue requires models to have robust generalization abilities and strong feature representation, aligning with the characteristics of large-scale pretrained models. However, large-scale models tend to lose representational ability in cross-domain scenarios due to limited sample diversity. \zlh{Given the abundant diversity provided by semantic modality, this paper leverages textual modality to enhance training sample diversity with CLP model}, meanwhile improving model transfer efficiency. Specifically, we propose the SeGD-VPT framework, which is divided into two phases. The first step aims to increase feature diversity by adding diversity prompts to each support sample, thereby generating varying input and enhancing sample diversity. Furthermore, we use diversity descriptions of classes to guide semantically meaningful learning of diversity prompts, proposing random combinations and selections of texts to increase textual diversity. Additionally, deep prompt tuning is introduced to enhance the model's transfer capability. After training of the first step, support samples with different diversity prompts are input into the CLIP backbone to generate enhanced features. After generation, the second phase trains classifiers using the generated features. Extensive experimental results across several benchmarks verify our method is comparable to SOTA source-utilized models and attain the best performance under the source-free CD-FSL setting.

Figures

Figures reproduced from arXiv: 2412.00767 by the authors.

Figure 1
Figure 1. Illustration of SeGD-VPT: 1) diversity prompts are [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Pipeline of the proposed SeGD-VPT framework. The SeGD-VPT employs a two-step training process. Step 1 aims to generate [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Accuracy curves demonstrate the impact of different [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: The t-SNE visualization results of our SeGD-VPT and CLIP model under 5-way 5-shot task cross four benchmarks. Different [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 1
Figure 1. Figure 1: Description examples from the "River" category in [PITH_FULL_IMAGE:figures/full_fig_p011_1.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. When Semantics Saturate or Emerge: Adaptation-Conditional Semantic Utility in Source-Free Cross-Domain Few-Shot Learning

    cs.CV 2026-08 conditional novelty 6.0 of 10

    Zero-shot prompt rankings do not predict post-adaptation usefulness: detailed descriptions saturate on EuroSAT and CropDisease but emerge on ISIC and ChestX.

  2. Multiple Stochastic Prompt Tuning for Few-shot Adaptation under Extreme Domain Shift

    cs.CV 2025-06 conditional novelty 5.0 of 10

    MIST adapts CLIP with two Gaussian-sampled prompts per class and improves few-shot accuracy under extreme domain shift on four benchmarks.

Reference graph

Works this paper leans on

61 extracted references · 43 canonical work pages · cited by 2 Pith papers

  1. [1]

    Dylan Auty and Krystian Mikolajczyk. 2023. Learning to Prompt CLIP for Monocular Depth Estimation: Exploring the Limits of Human Language. In Proceedings of the IEEE/CVF International Conference on Computer Vision . 2039–2047

    work page 2023

  2. [2]

    Shirsha Bose, Ankit Jha, Enrico Fini, Mainak Singha, Elisa Ricci, and Biplab Banerjee. 2024. Stylip: Multi-scale style-conditioned prompt learning for clip- based domain generalization. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 5542–5552

    work page 2024

  3. [3]

    Jingjing Chen, Linhai Zhuo, Zhipeng Wei, Hao Zhang, Huazhu Fu, and Yu-Gang Jiang. 2023. Knowledge driven weights estimation for large-scale few-shot image recognition. Pattern Recognition 142 (2023), 109668

    work page 2023

  4. [4]

    Junhyeong Cho, Gilhyun Nam, Sungyeon Kim, Hunmin Yang, and Suha Kwak

    work page

  5. [5]

    Noel Codella, Veronica Rotemberg, Philipp Tschandl, M Emre Celebi, Stephen Dusza, David Gutman, Brian Helba, Aadi Kalloo, Konstantinos Liopyris, Michael Marchetti, et al. 2019. Skin lesion analysis toward melanoma detection 2018: A challenge hosted by the international skin imaging collaboration (isic). arXiv preprint arXiv:1902.03368 (2019)

    arXiv 2019

  6. [6]

    Jiankang Deng, Jia Guo, Niannan Xue, and Stefanos Zafeiriou. 2019. Arcface: Additive angular margin loss for deep face recognition. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition. 4690–4699

    work page 2019

  7. [7]

    Yuqian Fu, Yanwei Fu, and Yu-Gang Jiang. 2021. Meta-FDMixup: Cross-Domain Few-Shot Learning Guided by Labeled Target Data. InProceedings of the 29th ACM International Conference on Multimedia. 5326–5334

    work page 2021

  8. [8]

    Yuqian Fu, Yu Xie, Yanwei Fu, Jingjing Chen, and Yu-Gang Jiang. 2022. ME- D2N: Multi-Expert Domain Decompositional Network for Cross-Domain Few- Shot Learning. In Proceedings of the 30th ACM International Conference on Multimedia. 6609–6617

    work page 2022

Show all 61 references
  1. [9]

    Yuqian Fu, Yu Xie, Yanwei Fu, Jingjing Chen, and Yu-Gang Jiang. 2022. Wave- SAN: Wavelet based Style Augmentation Network for Cross-Domain Few-Shot Learning. arXiv preprint arXiv:2203.07656 (2022)

    2022 arXiv

  2. [10]

    Yuqian Fu, Yu Xie, Yanwei Fu, and Yu-Gang Jiang. 2023. StyleAdv: Meta Style Adversarial Training for Cross-Domain Few-Shot Learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 24575– 24584

    2023

  3. [11]

    Peng Gao, Shijie Geng, Renrui Zhang, Teli Ma, Rongyao Fang, Yongfeng Zhang, Hongsheng Li, and Yu Qiao. 2021. Clip-adapter: Better vision-language models with feature adapters. arXiv preprint arXiv:2110.04544 (2021)

    2021 arXiv

  4. [12]

    Victor Garcia and Joan Bruna. 2017. Few-shot learning with graph neural networks. arXiv preprint arXiv:1711.04043 (2017)

    2017 arXiv

  5. [13]

    Sachin Goyal, Ananya Kumar, Sankalp Garg, Zico Kolter, and Aditi Raghunathan

  6. [14]

    Yunhui Guo, Noel C Codella, Leonid Karlinsky, James V Codella, John R Smith, Kate Saenko, Tajana Rosing, and Rogerio Feris. 2020. A broader study of cross- domain few-shot learning. In Computer Vision–ECCV 2020: 16th European Con- ference, Glasgow, UK, August 23–28, 2020, Proc...

    2020

  7. [15]

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

    Finetune like you pretrain: Improved finetuning of zero-shot vision models. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 19338–19347

  8. [16]

    Shell Xu Hu, Da Li, Jan Stühmer, Minyoung Kim, and Timothy M Hospedales

  9. [17]

    Patrick Helber, Benjamin Bischke, Andreas Dengel, and Damian Borth. 2019. Eurosat: A novel dataset and deep learning benchmark for land use and land cover classification. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing 12, 7 (2019), 2217–2226

    2019

  10. [18]

    Gabriel Ilharco, Mitchell Wortsman, Ross Wightman, Cade Gordon, Nicholas Carlini, Rohan Taori, Achal Dave, Vaishaal Shankar, Hongseok Namkoong, John Miller, Hannaneh Hajishirzi, Ali Farhadi, and Ludwig Schmidt. 2021. OpenCLIP. https://doi.org/10.5281/zenodo.5143773 If you use ...

    2021 doi

  11. [19]

    Ashraful Islam, Chun-Fu Richard Chen, Rameswar Panda, Leonid Karlinsky, Rogerio Feris, and Richard J Radke. 2021. Dynamic distillation network for cross- domain few-shot recognition with unlabeled data. Advances in Neural Information Processing Systems 34 (2021), 3584–3595

    2021

  12. [20]

    Yanxu Hu and Andy J Ma. 2022. Adversarial Feature Augmentation for Cross- domain Few-shot Classification. In Proceedings of the European Conference on Computer Vision (ECCV)

    2022

  13. [21]

    Brian Lester, Rami Al-Rfou, and Noah Constant. 2021. The power of scale for parameter-efficient prompt tuning. arXiv preprint arXiv:2104.08691 (2021)

    2021 arXiv

  14. [22]

    Aodi Li, Liansheng Zhuang, Shuo Fan, and Shafei Wang. 2022. Learning common and specific visual prompts for domain generalization. InProceedings of the Asian Conference on Computer Vision. 4260–4275

    2022

  15. [23]

    Dongjun Lee, Seokwon Song, Jihee Suh, Joonmyeong Choi, Sanghyeok Lee, and Hyunwoo J Kim. 2023. Read-only prompt optimization for vision-language few-shot learning. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 1401–1411

    2023

  16. [24]

    Xiang Lisa Li and Percy Liang. 2021. Prefix-tuning: Optimizing continuous prompts for generation. arXiv preprint arXiv:2101.00190 (2021)

    2021 arXiv

  17. [25]

    Hanwen Liang, Qiong Zhang, Peng Dai, and Juwei Lu. 2021. Boosting the Gen- eralization Capability in Cross-Domain Few-shot Learning via Noise-enhanced Supervised Autoencoder. In Proceedings of the IEEE/CVF International Confer- ence on Computer Vision. 9424–9434

    2021

  18. [26]

    Tianhong Li, Peng Cao, Yuan Yuan, Lijie Fan, Yuzhe Yang, Rogerio S Feris, Piotr Indyk, and Dina Katabi. 2022. Targeted supervised contrastive learning for long-tailed recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 6918–6928

    2022

  19. [27]

    Lingbo Liu, Jianlong Chang, Bruce XB Yu, Liang Lin, Qi Tian, and Chang- Wen Chen. 2022. Prompt-matched semantic segmentation. arXiv preprint arXiv:2208.10159 (2022)

    2022 arXiv

  20. [28]

    Pengfei Liu, Weizhe Yuan, Jinlan Fu, Zhengbao Jiang, Hiroaki Hayashi, and Graham Neubig. 2023. Pre-train, prompt, and predict: A systematic survey of prompting methods in natural language processing. Comput. Surveys 55, 9 (2023), 1–35

    2023

  21. [29]

    Bingyu Liu, Zhen Zhao, Zhenpeng Li, Jianan Jiang, Yuhong Guo, and Jieping Ye

  22. [30]

    Zuhao Liu, Xiao-Ming Wu, Dian Zheng, Kun-Yu Lin, and Wei-Shi Zheng. 2023. Generating Anomalies for Video Anomaly Detection With Prompt-Based Feature Mapping. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 24500–24510

    2023

  23. [31]

    Tianyi Ma, Yifan Sun, Zongxin Yang, and Yi Yang. 2023. ProD: Prompting-To- Disentangle Domain Knowledge for Cross-Domain Few-Shot Image Classification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 19754–19763

    2023

  24. [32]

    Atsuyuki Miyai, Qing Yu, Go Irie, and Kiyoharu Aizawa. 2023. LoCoOp: Few-Shot Out-of-Distribution Detection via Prompt Learning. arXiv preprint arXiv:2306.01293 (2023)

    2023 arXiv

  25. [33]

    Xiao Liu, Kaixuan Ji, Yicheng Fu, Weng Lam Tam, Zhengxiao Du, Zhilin Yang, and Jie Tang. 2021. P-tuning v2: Prompt tuning can be comparable to fine-tuning universally across scales and tasks. arXiv preprint arXiv:2110.07602 (2021)

    2021 arXiv

  26. [34]

    Cheng Perng Phoo and Bharath Hariharan. 2020. Self-training for few-shot transfer across extreme task differences. arXiv preprint arXiv:2010.07734 (2020)

    2020 arXiv

  27. [35]

    Zi Qian, Xin Wang, Xuguang Duan, Pengda Qin, Yuhong Li, and Wenwu Zhu

  28. [36]

    Jintao Rong, Hao Chen, Tianxiao Chen, Linlin Ou, Xinyi Yu, and Yifan Liu. 2023. Retrieval-Enhanced Visual Prompt Learning for Few-shot Classification. arXiv preprint arXiv:2306.02243 (2023)

    2023 arXiv

  29. [37]

    Sharada P Mohanty, David P Hughes, and Marcel Salathé. 2016. Using deep learning for image-based plant disease detection. Frontiers in plant science 7 (2016), 1419

    2016

  30. [38]

    Jiamei Sun, Sebastian Lapuschkin, Wojciech Samek, Yunqing Zhao, Ngai-Man Cheung, and Alexander Binder. 2021. Explanation-guided training for cross- domain few-shot classification. In 2020 25th International Conference on Pattern Recognition (ICPR). IEEE, 7609–7616

    2021

  31. [39]

    Philipp Tschandl, Cliff Rosendahl, and Harald Kittler. 2018. The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions. Scientific data 5, 1 (2018), 1–9

    2018

  32. [40]

    In Proceedings of the IEEE/CVF International Conference on Computer Vision

    Decouple before interact: Multi-modal prompt learning for continual visual question answering. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 2953–2962

  33. [41]

    Oriol Vinyals, Charles Blundell, Timothy Lillicrap, Daan Wierstra, et al. 2016. Matching networks for one shot learning. In Proceedings of the 30th Conference on Neural Information Processing Systems (NIPS). 3630–3638

    2016

  34. [42]

    Zhao Song, Ke Yang, Naiyang Guan, Junjie Zhu, Peng Qiao, and Qingyong Hu

  35. [43]

    In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP)

    VPPT: Visual Pre-Trained Prompt Tuning Framework for Few-Shot Image Classification. In ICASSP 2023-2023 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 1–5

    2023

  36. [44]

    Xiao Wang, Guangyao Chen, Guangwu Qian, Pengcheng Gao, Xiao-Yong Wei, Yaowei Wang, Yonghong Tian, and Wen Gao. 2023. Large-scale multi-modal pre-trained models: A comprehensive survey. Machine Intelligence Research (2023), 1–36

    2023

  37. [45]

    Xiaosong Wang, Yifan Peng, Le Lu, Zhiyong Lu, Mohammadhadi Bagheri, and Ronald M Summers. 2017. Chestx-ray8: Hospital-scale chest x-ray database ACM MM, 2024, Melbourne, Australia Linhai Zhuo, Zheng Wang, Tianwen Qian, and Y uqian Fu and benchmarks on weakly-supervised classif...

    2017

  38. [46]

    Hung-Yu Tseng, Hsin-Ying Lee, Jia-Bin Huang, and Ming-Hsuan Yang. 2020. Cross-domain few-shot classification via learned feature-wise transformation. arXiv preprint arXiv:2001.08735 (2020)

    2020 arXiv

  39. [47]

    Huali Xu, Li Liu, Shuaifeng Zhi, Shaojing Fu, Zhuo Su, Ming-Ming Cheng, and Yongxiang Liu. 2024. Enhancing Information Maximization with Distance-Aware Contrastive Learning for Source-Free Cross-Domain Few-Shot Learning. IEEE Transactions on Image Processing(2024)

    2024

  40. [49]

    Haoqing Wang and Zhi-Hong Deng. 2021. Cross-domain few-shot classification via adversarial task augmentation. arXiv preprint arXiv:2104.14385 (2021)

    2021 arXiv

  41. [50]

    Renrui Zhang, Rongyao Fang, Wei Zhang, Peng Gao, Kunchang Li, Jifeng Dai, Yu Qiao, and Hongsheng Li. 2021. Tip-adapter: Training-free clip-adapter for better vision-language modeling. arXiv preprint arXiv:2111.03930 (2021)

    2021 arXiv

  42. [51]

    Renrui Zhang, Xiangfei Hu, Bohao Li, Siyuan Huang, Hanqiu Deng, Yu Qiao, Peng Gao, and Hongsheng Li. 2023. Prompt, generate, then cache: Cascade of foundation models makes strong few-shot learners. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recogn...

    2023

  43. [52]

    Yong Wu, Shekhor Chanda, Mehrdad Hosseinzadeh, Zhi Liu, and Yang Wang

  44. [53]

    In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision

    Few-Shot Learning of Compact Models via Task-Specific Meta Distillation. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision. 6265–6274

  45. [54]

    Hao Zheng, Runqi Wang, Jianzhuang Liu, and Asako Kanezaki. 2023. Cross-level distillation and feature denoising for cross-domain few-shot classification. arXiv preprint arXiv:2311.02392 (2023)

    2023 arXiv

  46. [55]

    Liqi Yan, Cheng Han, Zenglin Xu, Dongfang Liu, and Qifan Wang. 2023. Prompt learns prompt: exploring knowledge-aware generative prompt collaboration for video captioning. In Proceedings of international joint conference on artificial intelligence (IJCAI). 1622–1630

    2023

  47. [56]

    Moslem Yazdanpanah and Parham Moradi. 2022. Visual domain bridge: A source- free domain adaptation for cross-domain few-shot learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2868–2877

    2022

  48. [59]

    Xin Zhang, Shixiang Shane Gu, Yutaka Matsuo, and Yusuke Iwasawa. 2023. Do- main prompt learning for efficiently adapting clip to unseen domains.Transactions of the Japanese Society for Artificial Intelligence 38, 6 (2023), B–MC2_1

    2023

  49. [60]

    Yifan Zhao, Tong Zhang, Jia Li, and Yonghong Tian. 2023. Dual adaptive repre- sentation alignment for cross-domain few-shot learning. IEEE Transactions on Pattern Analysis and Machine Intelligence (2023)

    2023

  50. [62]

    River" category in the EuroSAT dataset, as shown in Figure 1. We use a predefined template

    Linhai Zhuo, Yuqian Fu, Jingjing Chen, Yixin Cao, and Yu-Gang Jiang. 2022. TGDM: Target Guided Dynamic Mixup for Cross-Domain Few-Shot Learning. In Proceedings of the ACM International Conference on Multimedia (ACM MM). 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 ...

    2022 arXiv

  51. [2020]

    In arXiv preprint arXiv:2005.08463

    Feature transformation ensemble model with batch spectral regularization for cross-domain few-shot classification. In arXiv preprint arXiv:2005.08463

    2005 arXiv

  52. [2022]

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

    Pushing the Limits of Simple Pipelines for Few-Shot Learning: Exter- nal Data and Fine-Tuning Make a Difference. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 9068–9077

  53. [2023]

    In Proceedings of the IEEE/CVF International Conference on Computer Vision

    Promptstyler: Prompt-driven style generation for source-free domain gener- alization. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 15702–15712

Pith tools

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