Pith. sign in

REVIEW 3 major objections 5 minor 89 references

Modeling Multi-modal Cross-interaction for Multi-label Few-shot Image Classification Based on Local Feature Selection

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read Word embeddings can guide which image regions matter for each label, and using them to build prototypes makes multi-label few-shot image classification substantially more accurate.

desk verdict Solid extension of the authors' WGA line with useful new benchmarks, but the headline SOTA claim rests on a changed evaluation protocol; the only controlled comparison shows modest gains. read the letter →

arxiv 2412.13732 v2 pith:55SA7NJP submitted 2024-12-18 cs.CV

Kun Yan , Zied Bouraoui , Fangyun Wei , Chang Xu , Ping Wang , Shoaib Jameel , Steven Schockaert This is my paper · ORCID
classification cs.CV
keywords multi-labelfew-shotlearningprototyperefinementwordembeddingscross-modalattentiondynamicconvolutionlocalfeatureselectionlosschangemeasurementimageclassification
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 argues that in multi-label few-shot image classification, where each training image has several labels that refer to different image regions, label prototypes can be built reliably if word embeddings are used to first guess what each label means, then to select the image regions most likely to show that label, and finally to combine those regions with a cross-modal attention-and-convolution mechanism. The authors show that this three-stage prototype refinement substantially outperforms existing multi-label few-shot methods on COCO, PASCAL VOC, NUS-WIDE, and iMaterialist, and that it can even produce usable zero-shot-style predictions where prototypes come directly from word vectors. A reader should care because the approach shows a path to few-shot learning that relies on cheap textual priors instead of large labeled datasets, and because the reported gains are large enough to reset the state of the art on the LaSO benchmark.

What carries the argument

The load-bearing mechanism is the multi-modal cross-interaction module, composed of a channel-wise cross-attention in which the word-embedding prototype serves as the query and local visual features split by channel serve as keys and values, together with a word-embedding-based dynamic convolution that generates 1x1 convolutional kernels from the label embedding to modulate the selected local features. An optional Loss Change Measurement module precedes this, learning per-position importance weights and then using a first-order Taylor approximation of the loss change to hard-select the local features that matter. The joint embedding is learned with a cross-modality loss that treats projected word vectors as prototypes for classifying support images.

What would settle it

A test set of novel labels containing near-synonym pairs (for example 'sofa' and 'couch', or fine-grained iMaterialist attributes such as 'collar' and 'neckline') would falsify the central assumption if the attention maps and LCM selections fail to separate the corresponding regions, measured by chance-level pairwise discrimination accuracy or by no improvement over randomly selected local features.

Watch

Extended reading notes

Core claim

On the paper's own terms, the central discovery is that word embeddings, though too noisy to serve as final prototypes, carry enough semantic signal to guide every stage of prototype construction: a cross-modality loss aligns visual and textual features in a joint space, the word-embedding prototype acts as the query in a channel-wise cross-attention that aggregates local visual features, and the same embedding generates dynamic convolutional kernels that filter those features. A loss-change measurement module then hard-selects the local features whose removal would most change the classification loss, cleaning the support set before the soft attention step. The result is that on the LaSO multi-label few-shot benchmark, the proposed LCM model with a plain Conv-4-64 backbone achieves 60.3 mAP in 1-shot and 68.3 in 5-shot, beating methods that use the deeper GoogleNet-v3 backbone, and on COCO with a ResNet-50 backbone the model raises 1-shot Mi-AP from 35.30 (prior WGA) to 41.02 with GloVe and 43.20 with mirrorBERT.

Load-bearing premise

The method assumes that pre-trained word embeddings, after a learned linear projection, provide guidance accurate enough to select which local image regions belong to a label and to construct usable prototypes for novel labels, even when labels are visually or semantically close.

Editorial extensions

If this is right

  • On COCO, the LCM model with GloVe raises 1-shot Mi-AP from 35.30 (prior WGA) to 41.02, and with mirrorBERT to 43.20; the gains persist in 5-shot (from 38.97 to 46.35 with GloVe and 47.72 with mirrorBERT).
  • The method transfers to three newly proposed benchmarks (PASCAL VOC, NUS-WIDE, and iMaterialist) with consistent gains over prior state-of-the-art, suggesting the mechanism is not dataset-specific.
  • The base model can classify unseen labels without any fine-tuning, while the LCM variant requires only a short 20-epoch training step at test time.
  • Word embeddings from language models (such as mirrorBERT) outperform GloVe in every setting, and even zero-shot classification using the aligned embeddings shows mirrorBERT beating GloVe by 4.86 mAP.
  • Using CLIP image backbones further boosts performance: with ViT-B/32 the LCM model reaches 60.35 mAP in 1-shot and 65.01 in 5-shot on COCO.

Reading between the lines

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

  • The same three-stage prototype construction could be applied to other few-shot structured-prediction tasks where regions matter, such as few-shot object detection or part-based recognition, by swapping the region selector for a task-appropriate one.
  • The LCM hard-gating step is a separate test-time fine-tuning stage; a testable extension is to make the selection differentiable end-to-end (for example with Gumbel-Softmax), which might remove the 20-epoch training requirement.
  • The finding that mirrorBERT beats CLIP text embeddings for zero-shot prototype construction suggests that off-the-shelf language-model embeddings, aligned with a simple linear map, may be a cheaper and stronger prior than vision-language encoders for this task; that hypothesis goes beyond the paper's own conclusions.
  • The paper's own Figure 5 shows that semantically close labels (zebra vs cow) are not always separated, so a fair stress test would measure performance on fine-grained label sets, where the gains from word-embedding guidance may shrink.
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 / 5 minor

Summary. The paper addresses multi-label few-shot image classification (ML-FSIC) by constructing label prototypes from support-set local features, guided by pre-trained word embeddings. It proposes a base model with a cross-modality loss that aligns word and visual embeddings, word-embedding-guided channel-wise cross-attention, and word-embedding-based dynamic convolution; a variant additionally uses a Loss Change Measurement (LCM) module to select representative local features before prototype aggregation. The authors also introduce new ML-FSIC splits for PASCAL VOC, NUS-WIDE, and iMaterialist, and propose a new episodic evaluation protocol. Experiments on COCO and the new splits report large gains over reimplemented baselines and published results from prior work. Under the original LaSO evaluation protocol (Table 4), the method still improves over BCR but by smaller margins (e.g., 64.2 vs 61.5 mAP for 1-shot with GoogleNet-v3).

Significance. If the reported results hold under a controlled evaluation protocol, the paper makes a useful contribution to ML-FSIC: it demonstrates that word-embedding priors can be combined with local feature selection and multi-modal cross-interaction to build prototypes for novel labels, and it provides extensive ablations, robustness checks, and a zero-shot sanity check that partly supports the learned visual-text alignment. The code and data release are also valuable. However, the strength of the empirical claim is currently limited by protocol mismatches: the main tables use a newly proposed sampling protocol and include test-time adaptation for the LCM variant, while the only head-to-head comparison under the original LaSO protocol shows more modest gains over the existing state of the art.

major comments (3)
  1. [5.2, Tables 1–4] The central claim of a 'substantial margin' over the state of the art is not consistently supported by the evidence. Tables 1–3 use the authors' new support-set sampling protocol (all labels per episode, K images drawn independently, 200 episodes) and the LCM variant additionally runs 20 epochs of support-set optimization at test time (Section 5.1.2). The published numbers for KGGR, NLC, and BCR come from the original LaSO 10-episode protocol. The only controlled comparison under the original protocol is Table 4, where the LCM model with GoogleNet-v3 beats BCR by 2.7 mAP in 1-shot and 3.8 mAP in 5-shot, and the Conv-4-64 LCM model actually trails BCR in 1-shot (60.3 vs 61.5). The authors should either re-run the prior methods under the new protocol, restrict the 'substantial improvement' claim to the new protocol, or present Table 4 as the primary state-of-the-art comparison.
  2. [4.4, 5.1.2, Tables 1–4] The comparison is also confounded by an asymmetry in test-time computation and available priors. The LCM model performs 20 epochs of optimization on the support set at test time, while the base model and the published BCR/KGGR/NLC numbers do not. Additionally, the proposed method uses GloVe/contextualized word embeddings as an external prior, whereas BCR and NLC are reported without such a prior. The ablation in Table 5 shows that the CM loss is important, and Table 15 shows that the LCM selection helps, but these ablations are within the proposed pipeline. To establish that the multi-modal cross-interaction mechanism itself, rather than the test-time adaptation budget or the word-embedding prior, drives the gains, the authors should include a controlled ablation that applies the same test-time budget and word-embedding initialization to the strongest published baseline.
  3. [5.1.1, NUS-WIDE dataset] The NUS-WIDE split is built from images downloaded from a third-party Kaggle mirror because the official links expired. This creates a reproducibility risk: the exact image set, download date, and any preprocessing differences are not specified. The authors should provide the image IDs or a checksummed manifest, and verify that the labels from the original release match the downloaded images. Without this, the NUS-WIDE results in Tables 3 and 12 cannot be independently reproduced or compared with future work.
minor comments (5)
  1. [Section 1, contribution list] The contribution list contains a typo: 'New ML-FISC benchmarks' should read 'ML-FSIC benchmarks'.
  2. [Tables 1–3, 11–13] The main results tables do not report standard deviations or confidence intervals, even though Table 7 reports margins of error for COCO. Adding error bars for all main tables would help assess the stability of the reported gains, especially for the 1-shot setting.
  3. [4.2, Eqs. (8)–(11)] The first-order Taylor approximation used to estimate the loss change is introduced without a discussion of its accuracy. Since Table 15 shows that the Taylor approximation performs slightly better than directly using g(·), a brief explanation or empirical comparison of approximation error would clarify the reader's understanding.
  4. [5.5.2 and Figure 5] The paper honestly acknowledges that word embeddings are not always sufficiently informative, e.g., for distinguishing zebra and cow. It would strengthen the paper to quantify performance on novel labels with high semantic similarity, since this is precisely the regime where the central assumption of the method is under stress.
  5. [5.1.1] The NUS-WIDE Kaggle URL is split across lines in the typeset text and no version, license, or download date is given. Please provide a complete citation with access date and a persistent identifier.

Circularity Check

0 steps flagged · score 0.0 of 10

No equation-level circularity: prototype construction, LCM feature selection, and the zero-shot transfer test each have content beyond their inputs; the evaluation-protocol confounds are correctness risks, not circularity.

full rationale

The claimed derivation chain is not circular by construction. The CM loss (Eqs. 3-4) aligns external word vectors with visual features, but the final prototypes are visual aggregates produced by word-query cross-attention (Eqs. 12-14) and word-generated dynamic convolution (Eqs. 15-16), not the word vectors themselves; replacing word prototypes with support-feature prototypes is a genuine transformation. The LCM module (Eqs. 8-10) scores each local feature by the change in the alignment loss when that feature is ablated, which is an auxiliary criterion rather than a restatement of the query prediction (Eqs. 18-19). Word embeddings are pre-trained from text corpora and are not fitted to the novel-label test data, and Section 5.5.1 and Table 9 test the learned alignment directly in a zero-shot setting, providing an independent check. The paper's own limitation statement in Section 5.5.2 and Figure 5 acknowledge that word embeddings can be insufficiently informative for close label pairs, which is an honest restriction rather than a hidden identity. Self-citations are transparent: WGA [71] is presented as the prior base model being enhanced, [70,76] are cited as inspiration for LCM, and Table 4 compares against published external results (LaSO, KGGR, NLC, BCR) under the original setting. Concerns that the substantial margin in Tables 1-3 is confounded by the newly proposed sampling protocol, 200-episode averaging, and test-time LCM adaptation are evaluation-protocol questions, not cases where an equation reduces to its own inputs. No fitted parameter is renamed as a prediction, and no uniqueness theorem is imported from the authors' prior work.

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

The central claim rests on the transferability of word-visual alignment to novel labels and on the validity of gradient-based feature selection. No new physical or conceptual entities are proposed; the LCM module and attention mechanisms are algorithmic components rather than invented entities. The listed free parameters are tuned on validation or left unspecified, and each affects the reported results.

free parameters (6)
  • lambda (cosine scale) = not reported
    Scales the cosine similarity in the CM loss (Eq. 4) and query loss (Eq. 19). No value or tuning procedure is specified in the paper.
  • gamma (loss weight) = 1
    Controls the weight of the query loss relative to the CM loss in Eq. 20. Set to 1 based on the validation split (Section 5.1.3).
  • n_a (attention heads) = 8
    Number of attention heads in the cross-attention module. Tuned on the validation split (Section 5.1.3 and Fig. 3a).
  • theta (LCM threshold) = 0.65
    Sigmoid threshold that decides which local features are retained in the LCM module. Tuned on validation (Fig. 3b); theta=0.5 reduces to the base model.
  • n_d (dynamic conv selected features) = not specified
    Number of top local features fed into the word-embedding-based dynamic convolution (Eqs. 15-16). The value is never stated in the paper, an implementation gap.
  • d_j (joint embedding dimension) = 512
    Dimensionality of the shared embedding space for visual and text features. Set to 512 without an ablation or justification.
assumptions (4)
  • domain assumption Pre-trained word embeddings encode label semantics that can be linearly aligned with visual features.
    The entire prototype initialization and selection depend on this alignment; the paper uses GloVe, BERT, and CLIP embeddings without an analysis of per-label reliability. Figure 5 admits failures for semantically close labels.
  • ad hoc to paper The first-order Taylor approximation of the loss change sufficiently estimates local feature importance.
    Equation (10) replaces the exact loss change with rho times the gradient. This heuristic, borrowed from pruning literature, is used to select representative local features, but may not hold for all deep network losses.
  • domain assumption The linear alignment learned on base classes transfers to novel, unseen labels.
    The projection matrices A_visual and A_text are optimized only on base labels but are applied to novel labels at test time. Table 9 provides preliminary zero-shot evidence, but the transfer is not guaranteed.
  • domain assumption CNN local feature maps are spatially meaningful, so that individual grid cells correspond to objects or background.
    Local feature selection and attention operate on spatial grid positions. The paper provides qualitative visualizations but no quantitative spatial ground truth to validate this premise.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Modeling Multi-modal Cross-interaction for Multi-label Few-shot Image Classification Based on Local Feature Selection." pith.science (2026). https://pith.science/paper/55SA7NJP

@misc{pith2026241213732,
  author       = {Pith},
  title        = {Pith review of: Modeling Multi-modal Cross-interaction for Multi-label Few-shot Image Classification Based on Local Feature Selection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/55SA7NJP}},
  note         = {Machine review of arXiv:2412.13732}
}
read the original abstract

The aim of multi-label few-shot image classification (ML-FSIC) is to assign semantic labels to images, in settings where only a small number of training examples are available for each label. A key feature of the multi-label setting is that an image often has several labels, which typically refer to objects appearing in different regions of the image. When estimating label prototypes, in a metric-based setting, it is thus important to determine which regions are relevant for which labels, but the limited amount of training data and the noisy nature of local features make this highly challenging. As a solution, we propose a strategy in which label prototypes are gradually refined. First, we initialize the prototypes using word embeddings, which allows us to leverage prior knowledge about the meaning of the labels. Second, taking advantage of these initial prototypes, we then use a Loss Change Measurement (LCM) strategy to select the local features from the training images (i.e. the support set) that are most likely to be representative of a given label. Third, we construct the final prototype of the label by aggregating these representative local features using a multi-modal cross-interaction mechanism, which again relies on the initial word embedding-based prototypes. Experiments on COCO, PASCAL VOC, NUS-WIDE, and iMaterialist show that our model substantially improves the current state-of-the-art.

Figures

Figures reproduced from arXiv: 2412.13732 by the authors.

Figure 1
Figure 1. In the 1-shot single-label setting, a given training image can be interpreted as a prototype for the considered label (left). In the multi-label setting, labels are related to different regions of the image, and these regions need to be identified before meaningful prototypes can be obtained (right). ACM Reference Format: Kun Yan, Zied Bouraoui, Fangyun Wei, Chang Xu, Ping Wang, Shoaib Jameel, and Steven Schockaert.… view at source ↗
Figure 2
Figure 2. Overview of our methodology. A joint embedding space is learned in which both labels and images are represented. A customized multi-modal cross-interaction strategy is proposed to calculate label prototypes using local features from the relevant images in the support set, along with word vectors that provide prior knowledge about the considered label. In our visual representation, the green solid line indicates the … view at source ↗
Figure 3
Figure 3. Ablation study for the number of attention heads and threshold 𝜃. the word vectors produced by the text encoder of CLIP (obtained as explained in Section 5.1.3) exhibit weaker performance than GloVe and some of the BERT strategies. 5.4.5 Number of Attention Heads. In Fig. 3a, we analyze the importance of using multiple attention heads. It is important to note that adjustments to the number of attention heads necessi… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Visualization of attention weights among local features used to construct label prototypes for large objects. Darker areas indicate higher importance of the corresponding local features. mirrorBERT significantly outperforms both, with improvements of 4.86 and 5.63 mAP …
Figure 5
Figure 5. Figure 5: Visualization of the attention weights among local features for constructing label prototypes, using GloVe and mirrorBERT word vectors, across various categories. The color intensity indicates the relative importance of the corresponding local feature. features whose 𝜌…
Figure 6
Figure 6. Figure 6: Visualization of the distribution of importance scores among local features in the LCM module. The more important the local feature, the darker the corresponding visualization. J. ACM, Vol. 37, No. 4, Article 111. Publication date: August 2018 [PITH_FULL_IMAGE:figures…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

89 extracted references · 63 canonical work pages

  1. [1]

    Rabab Abdelfattah, Qing Guo, Xiaoguang Li, Xiaofeng Wang, and Song Wang. 2023. Cdul: Clip-driven unsupervised learning for multi-label image classification. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 1348–1357

    2023

  2. [2]

    Jean-Baptiste Alayrac, Jeff Donahue, Pauline Luc, Antoine Miech, Iain Barr, Yana Hasson, Karel Lenc, Arthur Mensch, Katherine Millican, Malcolm Reynolds, et al. 2022. Flamingo: a visual language model for few-shot learning. Advances in Neural Information Processing Systems 35 (2022), 23716–23736

    2022

  3. [3]

    Amit Alfassy, Leonid Karlinsky, Amit Aides, Joseph Shtok, Sivan Harary, Rogerio Feris, Raja Giryes, and Alex M Bronstein. 2019. Laso: Label-set operations networks for multi-label few-shot learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 6548–6557

    2019

  4. [4]

    Yuexuan An, Hui Xue, Xingyu Zhao, Ning Xu, Pengfei Fang, and Xin Geng. 2024. Leveraging Bilateral Correlations for Multi-Label Few-Shot Learning. IEEE Transactions on Neural Networks and Learning Systems (2024)

    2024

  5. [5]

    Piotr Bojanowski, Edouard Grave, Armand Joulin, and Tomas Mikolov. 2017. Enriching Word Vectors with Subword Information. Transactions of the Association for Computational Linguistics 5 (2017), 135–146. https://doi.org/10.1162/ tacl_a_00051

    2017

  6. [6]

    Shang-Fu Chen, Yi-Chen Chen, Chih-Kuan Yeh, and Yu-Chiang Wang. 2018. Order-free rnn with visual attention for multi-label classification. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 32

    2018

  7. [7]

    Tianshui Chen, Liang Lin, Xiaolu Hui, Riquan Chen, and Hefeng Wu. 2020. Knowledge-guided multi-label few-shot learning for general image recognition. IEEE Transactions on Pattern Analysis and Machine Intelligence (2020)

    2020

  8. [8]

    Zhengyu Chen, Jixie Ge, Heshen Zhan, Siteng Huang, and Donglin Wang. 2021. Pareto Self-Supervised Training for Few-Shot Learning. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 13663–13672

    2021

Show all 89 references
  1. [9]

    Zhao-Min Chen, Xiu-Shen Wei, Peng Wang, and Yanwen Guo. 2019. Multi-label image recognition with graph convolutional networks. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 5177– 5186

    2019

  2. [10]

    Hao Cheng, Yufei Wang, Haoliang Li, Alex C Kot, and Bihan Wen. 2023. Disentangled feature representation for few-shot image classification. IEEE Transactions on Neural Networks and Learning Systems (2023)

    2023

  3. [11]

    Xiao Chu, Wanli Ouyang, Hongsheng Li, and Xiaogang Wang. 2016. Structured feature learning for pose estimation. In Proceedings of the IEEE conference on computer vision and pattern recognition . 4715–4723

    2016

  4. [12]

    Tat-Seng Chua, Jinhui Tang, Richang Hong, Haojie Li, Zhiping Luo, and Yantao Zheng. 2009. NUS-WIDE: a real-world web image database from National University of Singapore. In Proceedings of the 8th ACM International Conference on Image and Video Retrieval, CIVR 2009, Santorini ...

    2009

  5. [13]

    Tat-Seng Chua, Jinhui Tang, Richang Hong, Haojie Li, Zhiping Luo, and Yantao Zheng. 2009. Nus-wide: a real-world web image database from national university of singapore. In Proceedings of the ACM international conference on image and video retrieval. 1–9

    2009

  6. [14]

    Elijah Cole, Oisin Mac Aodha, Titouan Lorieul, Pietro Perona, Dan Morris, and Nebojsa Jojic. 2021. Multi-label learning from single positive labels. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition . 933–942

    2021

  7. [15]

    Leichao Dai, Lin Feng, Xinglin Shang, and Han Su. 2023. Cross modal adaptive few-shot learning based on task dependence. Chinese Journal of Electronics 32, 1 (2023), 85–96

    2023

  8. [16]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2019. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. In Proceedings NAACL-HLT

    2019

  9. [17]

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

    2020 arXiv

  10. [18]

    Thibaut Durand, Nazanin Mehrasa, and Greg Mori. 2019. Learning a deep convnet for multi-label classification with partial labels. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 647–657

    2019

  11. [19]

    Mark Everingham, SM Ali Eslami, Luc Van Gool, Christopher KI Williams, John Winn, and Andrew Zisserman. 2015. The pascal visual object classes challenge: A retrospective. International journal of computer vision 111, 1 (2015), 98–136. J. ACM, Vol. 37, No. 4, Article 111. Publi...

    2015

  12. [20]

    Chelsea Finn, Pieter Abbeel, and Sergey Levine. 2017. Model-agnostic meta-learning for fast adaptation of deep networks. In International conference on machine learning . 1126–1135

    2017

  13. [21]

    Amit Gajbhiye, Luis Espinosa-Anke, and Steven Schockaert. 2022. Modelling commonsense properties using pre-trained bi-encoders. arXiv preprint arXiv:2210.02771 (2022)

    2022 arXiv

  14. [22]

    Peng Gao, Shijie Geng, Renrui Zhang, Teli Ma, Rongyao Fang, Yongfeng Zhang, Hongsheng Li, and Yu Qiao. 2024. Clip-adapter: Better vision-language models with feature adapters. International Journal of Computer Vision 132, 2 (2024), 581–595

    2024

  15. [23]

    Shijie Geng, Jianbo Yuan, Yu Tian, Yuxiao Chen, and Yongfeng Zhang. 2023. HiCLIP: Contrastive language-image pretraining with hierarchy-aware attention. arXiv preprint arXiv:2303.02995 (2023)

    2023 arXiv

  16. [24]

    Sheng Guo, Weilin Huang, Xiao Zhang, Prasanna Srikhanta, Yin Cui, Yuan Li, Hartwig Adam, Matthew R Scott, and Serge Belongie. 2019. The imaterialist fashion attribute dataset. In Proceedings of the IEEE/CVF International Conference on Computer Vision Workshops. 0–0

    2019

  17. [25]

    Zixian Guo, Bowen Dong, Zhilong Ji, Jinfeng Bai, Yiwen Guo, and Wangmeng Zuo. 2023. Texts as images in prompt tuning for multi-label image recognition. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 2808–2817

    2023

  18. [26]

    Kaiming He, Xiangyu Zhang, Shaoqing Ren, and Jian Sun. 2016. Deep residual learning for image recognition. In Proc. CVPR. 770–778

    2016

  19. [27]

    Chao Jia, Yinfei Yang, Ye Xia, Yi-Ting Chen, Zarana Parekh, Hieu Pham, Quoc Le, Yun-Hsuan Sung, Zhen Li, and Tom Duerig. 2021. Scaling up visual and vision-language representation learning with noisy text supervision. In International conference on machine learning . PMLR, 4904–4916

    2021

  20. [28]

    Jongmin Kim, Taesup Kim, Sungwoong Kim, and Chang D Yoo. 2019. Edge-labeling graph neural network for few-shot learning. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 11–20

    2019

  21. [29]

    Kipf and Max Welling

    Thomas N. Kipf and Max Welling. 2017. Semi-Supervised Classification with Graph Convolutional Networks. In 5th International Conference on Learning Representations, ICLR 2017, Toulon, France, April 24-26, 2017, Conference Track Proceedings

    2017

  22. [30]

    Wenlan Kuang and Zhixin Li. 2024. Multi-label image classification with multi-layered multi-perspective dynamic semantic representation. Machine Learning 113, 6 (2024), 3443–3461

    2024

  23. [31]

    Kibok Lee, Hao Yang, Satyaki Chakraborty, Zhaowei Cai, Gurumurthy Swaminathan, Avinash Ravichandran, and Onkar Dabeer. 2022. Rethinking few-shot object detection on a multi-domain benchmark. In European Conference on Computer Vision. 366–382

    2022

  24. [32]

    Aoxue Li, Weiran Huang, Xu Lan, Jiashi Feng, Zhenguo Li, and Liwei Wang. 2020. Boosting few-shot learning with adaptive margin loss. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 12576–12584

    2020

  25. [33]

    Junnan Li, Dongxu Li, Caiming Xiong, and Steven Hoi. 2022. Blip: Bootstrapping language-image pre-training for unified vision-language understanding and generation. In International Conference on Machine Learning . 12888–12900

    2022

  26. [34]

    Na Li, Zied Bouraoui, José Camacho-Collados, Luis Espinosa Anke, Qing Gu, and Steven Schockaert. 2021. Modelling General Properties of Nouns by Selectively Averaging Contextualised Embeddings. In Proceedings of the Thirtieth International Joint Conference on Artificial Intelli...

    2021

  27. [35]

    Na Li, Hanane Kteich, Zied Bouraoui, and Steven Schockaert. 2023. Distilling semantic concept embeddings from contrastively fine-tuned language models. arXiv preprint arXiv:2305.09785 (2023)

    2023 arXiv

  28. [36]

    Wenbin Li, Lei Wang, Jinglin Xu, Jing Huo, Yang Gao, and Jiebo Luo. 2019. Revisiting local descriptor based image- to-class measure for few-shot learning. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 7260–7268

    2019

  29. [37]

    Zhixin Li, Lan Lin, Canlong Zhang, Huifang Ma, Weizhong Zhao, and Zhiping Shi. 2021. A semi-supervised learning approach based on adaptive weighted fusion for automatic image annotation. ACM Transactions on Multimedia Computing, Communications, and Applications (TOMM) 17, 1 (2...

    2021

  30. [38]

    Zeqian Li, Michael Mozer, and Jacob Whitehill. 2021. Compositional embeddings for multi-label one-shot learning. In Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision . 296–304

    2021

  31. [39]

    Zhenguo Li, Fengwei Zhou, Fei Chen, and Hang Li. 2017. Meta-sgd: Learning to learn quickly for few-shot learning. arXiv preprint arXiv:1707.09835 (2017)

    2017 arXiv

  32. [40]

    Tsung-Yi Lin, Michael Maire, Serge Belongie, James Hays, Pietro Perona, Deva Ramanan, Piotr Dollár, and C Lawrence Zitnick. 2014. Microsoft coco: Common objects in context. In European conference on computer vision . Springer, 740–755

    2014

  33. [41]

    Fangyu Liu, Ivan Vulic, Anna Korhonen, and Nigel Collier. [n. d.]. Fast, Effective, and Self-Supervised: Transforming Masked Language Models into Universal Lexical and Sentence Encoders. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing,...

    2021

  34. [42]

    Qianchu Liu, Fangyu Liu, Nigel Collier, Anna Korhonen, and Ivan Vulic. 2021. MirrorWiC: On Eliciting Word-in- Context Representations from Pretrained Language Models. In Proceedings of the 25th Conference on Computational J. ACM, Vol. 37, No. 4, Article 111. Publication date: ...

    2021

  35. [43]

    Le, and Ilya Sutskever

    Tomás Mikolov, Quoc V. Le, and Ilya Sutskever. 2013. Exploiting Similarities among Languages for Machine Translation. CoRR abs/1309.4168 (2013). arXiv:1309.4168 http://arxiv.org/abs/1309.4168

    2013 arXiv

  36. [44]

    Tomas Mikolov, Wen-tau Yih, and Geoffrey Zweig. 2013. Linguistic Regularities in Continuous Space Word Represen- tations. In Proceedings NAACL-HLT. 746–751

    2013

  37. [45]

    Alex Nichol, Joshua Achiam, and John Schulman. 2018. On first-order meta-learning algorithms. arXiv preprint arXiv:1803.02999 (2018)

    2018 arXiv

  38. [46]

    Jeffrey Pennington, Richard Socher, and Christopher D Manning. 2014. Glove: Global vectors for word representation. In Proceedings of the 2014 conference on empirical methods in natural language processing (EMNLP) . 1532–1543

    2014

  39. [47]

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, et al. 2021. Learning transferable visual models from natural language supervision. In International conference on machine learni...

    2021

  40. [48]

    Sachin Ravi and Hugo Larochelle. 2017. Optimization as a model for few-shot learning. In International conference on learning representations

    2017

  41. [49]

    Rusu, Dushyant Rao, Jakub Sygnowski, Oriol Vinyals, Razvan Pascanu, Simon Osindero, and Raia Hadsell

    Andrei A. Rusu, Dushyant Rao, Jakub Sygnowski, Oriol Vinyals, Razvan Pascanu, Simon Osindero, and Raia Hadsell

  42. [50]

    Victor Garcia Satorras and Joan Bruna Estrach. 2018. Few-shot learning with graph neural networks. In International conference on learning representations

    2018

  43. [51]

    Julio Silva-Rodriguez, Sina Hajimiri, Ismail Ben Ayed, and Jose Dolz. 2023. A Closer Look at the Few-Shot Adaptation of Large Vision-Language Models. arXiv preprint arXiv:2312.12730 (2023)

    2023 arXiv

  44. [52]

    Christian Simon, Piotr Koniusz, and Mehrtash Harandi. 2022. Meta-learning for multi-label few-shot classification. In Proceedings of the IEEE/CVF winter conference on applications of computer vision . 3951–3960

    2022

  45. [53]

    Jake Snell, Kevin Swersky, and Richard Zemel. 2017. Prototypical networks for few-shot learning. Advances in neural information processing systems 30 (2017)

    2017

  46. [54]

    Peize Sun, Rufeng Zhang, Yi Jiang, Tao Kong, Chenfeng Xu, Wei Zhan, Masayoshi Tomizuka, Lei Li, Zehuan Yuan, Changhu Wang, et al. 2021. Sparse r-cnn: End-to-end object detection with learnable proposals. In Proceedings of the IEEE/CVF conference on computer vision and pattern ...

    2021

  47. [55]

    Ximeng Sun, Ping Hu, and Kate Saenko. 2022. Dualcoop: Fast adaptation to multi-label recognition with limited annotations. Advances in Neural Information Processing Systems 35 (2022), 30569–30582

    2022

  48. [56]

    Flood Sung, Yongxin Yang, Li Zhang, Tao Xiang, Philip HS Torr, and Timothy M Hospedales. 2018. Learning to compare: Relation network for few-shot learning. In Proceedings of the IEEE conference on computer vision and pattern recognition. 1199–1208

    2018

  49. [57]

    Christian Szegedy, Vincent Vanhoucke, Sergey Ioffe, Jon Shlens, and Zbigniew Wojna. 2016. Rethinking the inception architecture for computer vision. In Proceedings of the IEEE conference on computer vision and pattern recognition . 2818–2826

    2016

  50. [58]

    Antti Tarvainen and Harri Valpola. 2017. Mean teachers are better role models: Weight-averaged consistency targets improve semi-supervised deep learning results. Advances in neural information processing systems 30 (2017)

    2017

  51. [59]

    Grigorios Tsoumakas and Ioannis Katakis. 2007. Multi-label classification: An overview. International Journal of Data Warehousing and Mining (IJDWM) 3, 3 (2007), 1–13

    2007

  52. [60]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need. In Advances in neural information processing systems . 5998–6008

    2017

  53. [61]

    Oriol Vinyals, Charles Blundell, Tim Lillicrap, Koray Kavukcuoglu, and Daan Wierstra. 2016. Matching Networks for One Shot Learning. In Proc. NIPS. 3630–3638

    2016

  54. [62]

    Jiang Wang, Yi Yang, Junhua Mao, Zhiheng Huang, Chang Huang, and Wei Xu. 2016. Cnn-rnn: A unified framework for multi-label image classification. In Proceedings of the IEEE conference on computer vision and pattern recognition . 2285–2294

    2016

  55. [63]

    Yixiao Wang, Zied Bouraoui, Luis Espinosa Anke, and Steven Schockaert. 2022. Sentence Selection Strategies for Distilling Word Embeddings from BERT. InProceedings of the Thirteenth Language Resources and Evaluation Conference, LREC 2022, Marseille, France, 20-25 June 2022 . Eu...

    2022

  56. [64]

    Ya Wang, Dongliang He, Fu Li, Xiang Long, Zhichao Zhou, Jinwen Ma, and Shilei Wen. 2020. Multi-label classification with label graph superimposing. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 34. 12265–12272

    2020

  57. [65]

    Zhouxia Wang, Tianshui Chen, Guanbin Li, Ruijia Xu, and Liang Lin. 2017. Multi-label image recognition by recurrently discovering attentional regions. In Proceedings of the IEEE international conference on computer vision . 464–472

    2017

  58. [66]

    Zihao Wei, Zixuan Pan, and Andrew Owens. 2024. Efficient Vision-Language Pre-training by Cluster Masking. arXiv preprint arXiv:2405.08815 (2024). J. ACM, Vol. 37, No. 4, Article 111. Publication date: August 2018. Modeling Multi-modal Cross-interaction for ML-FSIC Based on Loc...

    2024 arXiv

  59. [67]

    Chen Xing, Negar Rostamzadeh, Boris Oreshkin, and Pedro O O Pinheiro. 2019. Adaptive cross-modal few-shot learning. Advances in Neural Information Processing Systems 32 (2019)

    2019

  60. [68]

    Kun Yan, Zied Bouraoui, Ping Wang, Shoaib Jameel, and Steven Schockaert. 2021. Aligning visual prototypes with BERT embeddings for few-shot learning. In Proceedings of the 2021 International Conference on Multimedia Retrieval . 367–375

    2021

  61. [69]

    Kun Yan, Zied Bouraoui, Ping Wang, Shoaib Jameel, and Steven Schockaert. 2021. Few-shot image classification with multi-facet prototypes. In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP). IEEE, 1740–1744

    2021

  62. [70]

    Kun Yan, Lingbo Liu, Jun Hou, and Ping Wang. 2021. Representative Local Feature Mining for Few-Shot Learning. In ICASSP 2021-2021 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 1730–1734

    2021

  63. [71]

    Kun Yan, Chenbin Zhang, Jun Hou, Ping Wang, Zied Bouraoui, Shoaib Jameel, and Steven Schockaert. 2022. Inferring prototypes for multi-label few-shot image classification with word vector guided attention. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 3...

    2022

  64. [72]

    Kaicheng Yang, Jiankang Deng, Xiang An, Jiawei Li, Ziyong Feng, Jia Guo, Jing Yang, and Tongliang Liu. 2023. Alip: Adaptive language-image pre-training with synthetic caption. In Proceedings of the IEEE/CVF International Conference on Computer Vision. 2922–2931

    2023

  65. [73]

    Vacit Oguz Yazici, Abel Gonzalez-Garcia, Arnau Ramisa, Bartlomiej Twardowski, and Joost van de Weijer. 2020. Orderless recurrent models for multi-label classification. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 13440–13449

    2020

  66. [74]

    Han-Jia Ye, Hexiang Hu, De-Chuan Zhan, and Fei Sha. 2020. Few-shot learning via embedding adaptation with set-to-set functions. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 8808–8817

    2020

  67. [75]

    Renchun You, Zhiyao Guo, Lei Cui, Xiang Long, Yingze Bao, and Shilei Wen. 2020. Cross-modality attention with semantic graph embedding for multi-label classification. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 34. 12709–12716

    2020

  68. [76]

    Zhonghui You, Kun Yan, Jinmian Ye, Meng Ma, and Ping Wang. 2019. Gate decorator: Global filter pruning method for accelerating deep convolutional neural networks. Advances in Neural Information Processing Systems 32 (2019)

    2019

  69. [77]

    Jin Yuan, Shikai Chen, Yao Zhang, Zhongchao Shi, Xin Geng, Jianping Fan, and Yong Rui. 2023. Graph attention transformer network for multi-label image classification.ACM Transactions on Multimedia Computing, Communications and Applications 19, 4 (2023), 1–16

    2023

  70. [78]

    Chi Zhang, Yujun Cai, Guosheng Lin, and Chunhua Shen. 2022. Deepemd: Differentiable earth mover’s distance for few-shot learning. IEEE Transactions on Pattern Analysis and Machine Intelligence (2022)

    2022

  71. [79]

    Ruru Zhang, E Haihong, Meina Song, and Xun Cao. 2024. FSCIL-EACA: Few-Shot Class-Incremental Learning Network Based on Embedding Augmentation and Classifier Adaptation for Image Classification. Chinese Journal of Electronics 33, 1 (2024), 139–152

    2024

  72. [80]

    Yunfeng Zhao, Guoxian Yu, Lei Liu, Zhongmin Yan, Carlotta Domeniconi, and Lizhen Cui. 2021. Few-shot partial multi-label learning. In 2021 IEEE International Conference on Data Mining (ICDM) . IEEE, 926–935

    2021

  73. [81]

    Yunfeng Zhao, Guoxian Yu, Lei Liu, Zhongmin Yan, Carlotta Domeniconi, Xiayan Zhang, and Lizhen Cui. 2023. Few-shot partial multi-label learning via prototype rectification. Knowledge and Information Systems 65, 4 (2023), 1851–1880

    2023

  74. [82]

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. 2022. Conditional prompt learning for vision-language models. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition . 16816–16825

    2022

  75. [83]

    Kaiyang Zhou, Jingkang Yang, Chen Change Loy, and Ziwei Liu. 2022. Learning to prompt for vision-language models. International Journal of Computer Vision 130, 9 (2022), 2337–2348

    2022

  76. [84]

    Wei Zhou, Peng Dou, Tao Su, Haifeng Hu, and Zhijie Zheng. 2023. Feature learning network with transformer for multi-label image classification. Pattern Recognition 136 (2023), 109203

    2023

  77. [85]

    Wei Zhou, Yanke Hou, Dihu Chen, Haifeng Hu, and Tao Su. 2023. Attention-augmented memory network for image multi-label classification. ACM Transactions on Multimedia Computing, Communications and Applications 19, 3 (2023), 1–24

    2023

  78. [86]

    Wei Zhou, Zhiwu Xia, Peng Dou, Tao Su, and Haifeng Hu. 2023. Double attention based on graph attention network for image multi-label classification. ACM Transactions on Multimedia Computing, Communications and Applications 19, 1 (2023), 1–23

    2023

  79. [87]

    Feng Zhu, Hongsheng Li, Wanli Ouyang, Nenghai Yu, and Xiaogang Wang. 2017. Learning spatial regularization with image-level supervisions for multi-label image classification. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition. 5513–5522

    2017

  80. [88]

    Ke Zhu and Jianxin Wu. 2021. Residual attention: A simple but effective method for multi-label recognition. In Proceedings of the IEEE/CVF International Conference on Computer Vision . 184–193. J. ACM, Vol. 37, No. 4, Article 111. Publication date: August 2018. 111:24 Kun et a...

    2021

  81. [2019]

    In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019

    Meta-Learning with Latent Embedding Optimization. In 7th International Conference on Learning Representations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019

    2019

Pith tools

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