pith. sign in

arxiv: 2510.18326 · v3 · pith:7TNGYTQLnew · submitted 2025-10-21 · 💻 cs.CV

Enhancing Few-Shot Classification of Benchmark and Disaster Imagery with ABHFA-Net

Gao Yu Lee , Tanmoy Dam , Md Meftahul Ferdaus , Daniel Puiu Poenar , Vu Duong This is my paper

Pith reviewed 2026-05-18 05:16 UTC · model grok-4.3

classification 💻 cs.CV
keywords few-shot learningBhattacharyya distanceattention mechanismdisaster image classificationremote sensingprototype learningcontrastive lossimage classification
0
0 comments X

The pith

ABHFA-Net models class prototypes as probability distributions and classifies via Bhattacharyya distance to boost few-shot accuracy on disaster and benchmark images.

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

The paper presents ABHFA-Net as a few-shot framework that combines spatial-channel attention with a Bhattacharyya-based contrastive softmax loss to improve feature discrimination when labeled data is scarce. The method targets remote-sensing disaster imagery that shows high intra-class variation and limited samples. A sympathetic reader would see value in any approach that delivers reliable classification without requiring large annotated sets for time-sensitive applications such as emergency response. Experiments report gains over prior methods on both standard benchmarks and actual disaster collections.

Core claim

ABHFA-Net models class prototypes as probability distributions and performs classification via Bhattacharyya distance-based comparison. It integrates a spatial-channel attention mechanism to enhance discriminative feature learning in the few-shot context and introduces a Bhattacharyya-based contrastive softmax loss for improved class separability. On CIFAR-FS the network reaches 80.7 percent accuracy in the 5-way 1-shot setting and 92.3 percent in the 5-shot setting while also raising performance on the AIDER disaster set to 68.2 percent (1-shot) and 78.3 percent (5-shot).

What carries the argument

ABHFA-Net, the Attention Bhattacharyya Distance-based Feature Aggregation Network that represents prototypes as distributions and compares them with Bhattacharyya distance while applying attention to refine features.

If this is right

  • Outperforms prior state-of-the-art methods on CIFAR-FS, FC-100, miniImageNet and tieredImageNet under standard 5-way 1-shot and 5-shot protocols.
  • Raises 1-shot and 5-shot accuracy on real disaster collections including AIDER, CDD and MEDIC.
  • Supplies a practical model for data-scarce remote-sensing tasks where rapid deployment is required.
  • Demonstrates that distribution-based prototype comparison plus attention yields measurable class-separability gains in high-variability imagery.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same distribution-modeling and Bhattacharyya comparison could be tested on other variable imagery domains such as medical scans or agricultural monitoring.
  • Pairing the loss with different backbone architectures or unsupervised pre-training might further reduce the labeled-sample requirement.
  • The method's emphasis on separability under scarcity suggests it may complement existing metric-learning techniques rather than replace them outright.

Load-bearing premise

The reported accuracy gains arise from the specific pairing of spatial-channel attention and Bhattacharyya contrastive loss rather than from dataset-specific tuning or evaluation-protocol choices.

What would settle it

An independent re-run of the baselines using identical hyper-parameter search budgets and random seeds that produces no statistically significant accuracy difference on the same splits would falsify the claim of genuine improvement from the new components.

read the original abstract

The rising incidence of natural and human-induced disasters necessitates robust visual recognition systems capable of operating under limited labeled data conditions. However, disaster-related image classification remains challenging due to data scarcity, high intra-class variability, and domain-specific complexities in remote sensing imagery. To address these challenges, we propose the Attention Bhattacharyya Distance-based Feature Aggregation Network (ABHFA-Net), a novel few-shot learning (FSL) framework that models class prototypes as probability distributions and performs classification via Bhattacharyya distance-based comparison. Our approach integrates a spatial channel attention mechanism to enhance discrimiantive feature learning in the few-shot context and introduces a Bhattacharyya-based contrastive softmax loss for improved class separability. Extensive experiments on both benchmark datasets (CIFAR-FS, FC-100, miniImageNet, tieredImageNet) and real-world disaster datasets (AIDER, CDD, MEDIC) demonstrate the effectiveness of the proposed method. In particular, ABHFA-Net achieves 80.7% and 92.3% accuracy on CIFAR-FS under 5-way 1-shot and 5-shot settings, respectively, outperforming existing state-of-the-art methods. On disaster datasets, the model consistently improves classification performance, achieving up to 68.2% (1-shot) and 78.3% (5-shot) accuracy on AIDER, highlighting its robustness in real-world scenarios. These results establish ABHFA-Net as a strong and practical solution for few-shot disaster image classification, particularly in data-scarce and time-critical remote sensing applications. The code repository for our work is available at https://github.com/GreedYLearner1146/ABHFA-Net.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper proposes ABHFA-Net, a few-shot learning architecture for image classification that models class prototypes as probability distributions, incorporates a spatial-channel attention mechanism, and employs a Bhattacharyya distance-based contrastive softmax loss to improve discriminative power and separability. It evaluates the method on standard benchmarks (CIFAR-FS, FC-100, miniImageNet, tieredImageNet) and disaster-related datasets (AIDER, CDD, MEDIC), reporting state-of-the-art results including 80.7% accuracy on CIFAR-FS 5-way 1-shot and 92.3% on 5-shot, as well as gains up to 68.2% (1-shot) and 78.3% (5-shot) on AIDER. The code is made publicly available.

Significance. If the performance improvements can be rigorously attributed to the attention and Bhattacharyya components rather than hyperparameter choices, the work would provide a useful empirical contribution to few-shot classification in data-scarce remote sensing and disaster response scenarios. The public code repository is a clear strength that aids reproducibility.

major comments (2)
  1. [Section 4] Section 4 (Experiments and Results): The reported accuracies (e.g., 80.7% and 92.3% on CIFAR-FS, 68.2% and 78.3% on AIDER) are given as single point estimates without standard deviations, results over multiple random seeds, or explicit fixed data splits and statistical testing, which weakens the central claim of consistent outperformance over prior methods.
  2. [Section 3] Section 3 (Method): No ablation tables or controlled experiments are presented that isolate the spatial-channel attention module or replace the Bhattacharyya-based contrastive softmax loss with a standard cross-entropy or prototypical loss while holding backbone, optimizer, episode sampling, and all other factors fixed; this leaves open the possibility that gains arise from dataset-specific tuning of the free parameters (attention dimensions and loss weights).
minor comments (2)
  1. [Abstract] Abstract: Typo in 'discrimiantive' (should be 'discriminative').
  2. [Throughout] Throughout: More explicit description of the exact episode sampling protocol, number of ways/shots per episode, and hyperparameter search procedure would improve clarity and allow better assessment of the experimental protocol.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and the recommendation for major revision. We address each major comment below and will update the manuscript to improve experimental rigor.

read point-by-point responses

  1. Referee: [Section 4] Section 4 (Experiments and Results): The reported accuracies (e.g., 80.7% and 92.3% on CIFAR-FS, 68.2% and 78.3% on AIDER) are given as single point estimates without standard deviations, results over multiple random seeds, or explicit fixed data splits and statistical testing, which weakens the central claim of consistent outperformance over prior methods.

    Authors: We agree that single-point estimates without variability measures or statistical support weaken the claims of consistent outperformance. In the revised manuscript we will rerun all experiments over multiple random seeds (reporting means and standard deviations), explicitly document the fixed data splits, and include statistical significance tests (e.g., paired t-tests) against the baselines. These additions will appear in an updated Section 4. revision: yes

  2. Referee: [Section 3] Section 3 (Method): No ablation tables or controlled experiments are presented that isolate the spatial-channel attention module or replace the Bhattacharyya-based contrastive softmax loss with a standard cross-entropy or prototypical loss while holding backbone, optimizer, episode sampling, and all other factors fixed; this leaves open the possibility that gains arise from dataset-specific tuning of the free parameters (attention dimensions and loss weights).

    Authors: We acknowledge that the absence of controlled ablations leaves the source of gains ambiguous. We will add a dedicated ablation study in the revision that removes or substitutes the spatial-channel attention module and replaces the Bhattacharyya contrastive softmax loss with cross-entropy or standard prototypical loss, while freezing the backbone, optimizer, episode sampling, and all other hyperparameters. Results will be presented in a new table to isolate each component's contribution. revision: yes

Circularity Check

0 steps flagged

No significant circularity: empirical results on public benchmarks are externally verifiable

full rationale

The manuscript proposes ABHFA-Net as an architectural combination of spatial-channel attention and a Bhattacharyya-based contrastive softmax loss for few-shot image classification. All load-bearing claims are empirical accuracies reported on standard public datasets (CIFAR-FS, miniImageNet, tieredImageNet, AIDER, etc.) under conventional 5-way 1-shot and 5-shot protocols. No mathematical derivation, uniqueness theorem, or closed-form prediction is presented that reduces by construction to fitted constants, self-citations, or renamed inputs. The evaluation protocol, while lacking detailed ablations in the provided text, operates on externally reproducible benchmarks and does not embed the target performance numbers inside the model definition or loss formulation. This is a standard empirical ML contribution whose validity can be checked by re-running the released code on the cited datasets; therefore the derivation chain contains no circular steps.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 1 invented entities

The method relies on standard few-shot learning assumptions and a set of network hyperparameters that are tuned on the reported benchmarks.

free parameters (1)
  • attention module dimensions and loss weighting coefficients
    These control the strength of the spatial-channel attention and the contrastive term and are selected to maximize reported accuracies.
axioms (1)
  • domain assumption Class prototypes can be usefully modeled as probability distributions rather than point estimates in few-shot regimes
    Invoked when the paper states that modeling prototypes as distributions enables Bhattacharyya distance comparison.
invented entities (1)
  • ABHFA-Net no independent evidence
    purpose: A new network that aggregates features via attention and Bhattacharyya distance for few-shot disaster classification
    Newly proposed architecture whose performance is demonstrated only through the experiments in this paper.

pith-pipeline@v0.9.0 · 5861 in / 1399 out tokens · 37683 ms · 2026-05-18T05:16:21.312655+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

96 extracted references · 96 canonical work pages · 5 internal anchors

  1. [1]

    In: 2020 4t h International Conference on Trends in Electronics and Informatics (ICOEI)(48 184), pp

    Valsan, A., Parvathy, B., GH, V.D., Unnikrishnan, R., Reddy, P.K., Viv ek, A.: Unmanned aerial vehicle for search and rescue mission. In: 2020 4t h International Conference on Trends in Electronics and Informatics (ICOEI)(48 184), pp. 684– 687 (2020). IEEE 40

    work page 2020

  2. [2]

    IEEE Journ al of Selected Topics in Applied Earth Observations and Remote Sensing 14, 2387–2402 (2021)

    Sun, X., Wang, B., Wang, Z., Li, H., Li, H., Fu, K.: Research progress on few- shot learning for remote sensing image interpretation. IEEE Journ al of Selected Topics in Applied Earth Observations and Remote Sensing 14, 2387–2402 (2021)

    work page 2021

  3. [3]

    Ar tificial Intelligence Review 57(7), 169 (2024)

    Lee, G.Y., Dam, T., Ferdaus, M.M., Poenar, D.P., Duong, V.N.: Unlockin g the capabilities of explainable few-shot learning in remote sensing. Ar tificial Intelligence Review 57(7), 169 (2024)

    work page 2024

  4. [4]

    Nature 448(7153), 575–578 (2007)

    Ramanathan, V., Ramana, M.V., Roberts, G., Kim, D., Corrigan, C., C hung, C., Winker, D.: Warming trends in asia amplified by brown cloud solar absorp tion. Nature 448(7153), 575–578 (2007)

    work page 2007

  5. [5]

    IEEE Transactions on Geoscience and Remote Sensing 59(3), 1868–1875 (2020)

    Tan, A.E.-C., McCulloch, J., Rack, W., Platt, I., Woodhead, I.: Radar mea- surements of snow depth over sea ice on an unmanned aerial vehicle . IEEE Transactions on Geoscience and Remote Sensing 59(3), 1868–1875 (2020)

    work page 2020

  6. [6]

    : A novel disaster image data-set and characteristics analysis using attention model

    Niloy, F.F., Nayem, A.B.S., Sarker, A., Paul, O., Amin, M.A., Ali, A.A., Zaber , M.I., Rahman, A.M., et al. : A novel disaster image data-set and characteristics analysis using attention model. In: 2020 25th International Confe rence on Pattern Recognition (ICPR), pp. 6116–6122 (2021). IEEE

    work page 2020

  7. [7]

    IEEE Journal of Selected Topics in Applied Earth Observations and R emote Sensing 13, 1687–1699 (2020)

    Kyrkou, C., Theocharides, T.: Emergencynet: Efficient aerial ima ge classification for drone-based emergency monitoring using atrous convolutiona l feature fusion. IEEE Journal of Selected Topics in Applied Earth Observations and R emote Sensing 13, 1687–1699 (2020)

    work page 2020

  8. [8]

    IEEE Tran sactions on Geoscience and Remote Sensing 57(9), 6690–6709 (2019)

    Li, S., Song, W., Fang, L., Chen, Y., Ghamisi, P., Benediktsson, J.A.: D eep learn- ing for hyperspectral image classification: An overview. IEEE Tran sactions on Geoscience and Remote Sensing 57(9), 6690–6709 (2019)

    work page 2019

  9. [9]

    IEEE Geoscience and Remote Sensing Letters (2023)

    Lee, G.Y., Dam, T., Ferdaus, M.M., Poenar, D.P., Duong, V.N.: Watt-e ffnet: A lightweight and accurate model for classifying aerial disaster imag es. IEEE Geoscience and Remote Sensing Letters (2023)

    work page 2023

  10. [10]

    arXiv preprint arXiv:1904.04232 (2019)

    Chen, W.-Y., Liu, Y.-C., Kira, Z., Wang, Y.-C.F., Huang, J.-B.: A closer look at few-shot classification. arXiv preprint arXiv:1904.04232 (2019)

    work page arXiv 1904

  11. [11]

    In: Proceedings of the IEEE/CVF Conference on Comput er Vision and Pattern Recognition, pp

    Simon, C., Koniusz, P., Nock, R., Harandi, M.: Adaptive subspaces for few-shot learning. In: Proceedings of the IEEE/CVF Conference on Comput er Vision and Pattern Recognition, pp. 4136–4145 (2020)

    work page 2020

  12. [12]

    : Siamese neural networks for one- shot image recognition

    Koch, G., Zemel, R., Salakhutdinov, R., et al. : Siamese neural networks for one- shot image recognition. In: ICML Deep Learning Workshop, vol. 2 (2 015). Lille

    work page

  13. [13]

    In : Similarity- Based Pattern Recognition: Third International Workshop, SIMB AD 2015, Copenhagen, Denmark, October 12-14, 2015

    Hoffer, E., Ailon, N.: Deep metric learning using triplet network. In : Similarity- Based Pattern Recognition: Third International Workshop, SIMB AD 2015, Copenhagen, Denmark, October 12-14, 2015. Proceedings 3, pp . 84–92 (2015). 41 Springer

    work page 2015

  14. [14]

    Advances in neural information processing systems 30 (2017)

    Snell, J., Swersky, K., Zemel, R.: Prototypical networks for few -shot learning. Advances in neural information processing systems 30 (2017)

    work page 2017

  15. [15]

    IEEE T ransactions on Geoscience and Remote Sensing (2023)

    Zhang, M., Liu, H., Gong, M., Li, H., Wu, Y., Jiang, X.: Cross-domain s elf-taught network for few-shot hyperspectral image classification. IEEE T ransactions on Geoscience and Remote Sensing (2023)

    work page 2023

  16. [16]

    Science 350(6266), 1332–1338 (2015)

    Lake, B.M., Salakhutdinov, R., Tenenbaum, J.B.: Human-level con cept learning through probabilistic program induction. Science 350(6266), 1332–1338 (2015)

    work page 2015

  17. [17]

    Advances in neural information processing syst ems 29 (2016)

    Vinyals, O., Blundell, C., Lillicrap, T., Wierstra, D., et al.: Matching net works for one shot learning. Advances in neural information processing syst ems 29 (2016)

    work page 2016

  18. [18]

    Meta-Learning for Semi-Supervised Few-Shot Classification

    Ren, M., Triantafillou, E., Ravi, S., Snell, J., Swersky, K., Tenenbau m, J.B., Larochelle, H., Zemel, R.S.: Meta-learning for semi-supervised few-s hot classifi- cation. arXiv preprint arXiv:1803.00676 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  19. [19]

    Advances in neural informa tion processing systems 31 (2018)

    Oreshkin, B., Rodr ´ ıguez L´ opez, P., Lacoste, A.: Tadam: Taskdependent adaptive metric for improved few-shot learning. Advances in neural informa tion processing systems 31 (2018)

    work page 2018

  20. [20]

    Krizhevsky, A., Hinton, G., et al.: Learning multiple layers of featu res from tiny images (2009)

    work page 2009

  21. [21]

    Journal of Engineering a nd Applied Science 70(1), 1–14 (2023)

    Meng, X., Wang, X., Yin, S., Li, H.: Few-shot image classification algo rithm based on attention mechanism and weight fusion. Journal of Engineering a nd Applied Science 70(1), 1–14 (2023)

    work page 2023

  22. [22]

    arXiv preprint arXiv:2008.02465 (2020)

    Jiang, Z., Kang, B., Zhou, K., Feng, J.: Few-shot classification via adaptive attention. arXiv preprint arXiv:2008.02465 (2020)

    work page arXiv 2008

  23. [23]

    Pattern Recognition 135, 109170 (2023)

    Huang, X., Choi, S.H.: Sapenet: Self-attention based prototyp e enhancement network for few-shot learning. Pattern Recognition 135, 109170 (2023)

    work page 2023

  24. [24]

    Neurocomputing 466, 16–26 (2021)

    Wang, C., Song, S., Yang, Q., Li, X., Huang, G.: Fine-grained few sh ot learning with foreground object transformation. Neurocomputing 466, 16–26 (2021)

    work page 2021

  25. [25]

    arXiv e-prints, 2310 (2023)

    Zhang, J., Luo, X., Gao, L., Zou, D., Shen, H., Song, J.: From chan nel bias to feature redundancy: Uncovering the” less is more” principle in few- shot learning. arXiv e-prints, 2310 (2023)

    work page 2023

  26. [26]

    In: 2023 IEEE 35th International Confere nce on Tools with Artificial Intelligence (ICTAI), pp

    Shangguan, Z., Huai, L., Liu, T., Jiang, X.: Few-shot object dete ction with refined contrastive learning. In: 2023 IEEE 35th International Confere nce on Tools with Artificial Intelligence (ICTAI), pp. 991–996 (2023). IEEE

    work page 2023

  27. [27]

    arXiv p reprint arXiv:2403.13375 (2024)

    Zhou, J., Li, W., Cao, Y., Cai, H., Li, X.: Few-shot oriented object d etection 42 with memorable contrastive learning in remote sensing images. arXiv p reprint arXiv:2403.13375 (2024)

    work page arXiv 2024

  28. [28]

    arXiv preprint arXiv:2301.13411 (2023)

    Han, J., Ren, Y., Ding, J., Yan, K., Xia, G.-S.: Few-shot object det ection via variational feature aggregation. arXiv preprint arXiv:2301.13411 (2023)

    work page arXiv 2023

  29. [29]

    In: Proceedings of the IEEE/CVF International Conference on Comp uter Vision, pp

    Zhang, J., Zhao, C., Ni, B., Xu, M., Yang, X.: Variational few-shot learning. In: Proceedings of the IEEE/CVF International Conference on Comp uter Vision, pp. 1685–1694 (2019)

    work page 2019

  30. [30]

    Applied Intelligence 54(2), 1879–1892 (2024)

    Pan, M., Shen, H.: Multimodal variational contrastive learning fo r few-shot classification. Applied Intelligence 54(2), 1879–1892 (2024)

    work page 2024

  31. [31]

    arXiv preprint arXiv:1711.01558 (2017)

    Tolstikhin, I., Bousquet, O., Gelly, S., Schoelkopf, B.: Wasserste in auto-encoders. arXiv preprint arXiv:1711.01558 (2017)

    work page arXiv 2017

  32. [32]

    Gaussian Prototypical Networks for Few-Shot Learning on Omniglot

    Fort, S.: Gaussian prototypical networks for few-shot learn ing on omniglot. arXiv preprint arXiv:1708.02735 (2017)

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  33. [33]

    In: Proceedings of the IEEE/CVF Conference on Co mputer Vision and Pattern Recognition, pp

    Bateni, P., Goyal, R., Masrani, V., Wood, F., Sigal, L.: Improved few -shot visual classification. In: Proceedings of the IEEE/CVF Conference on Co mputer Vision and Pattern Recognition, pp. 14493–14502 (2020)

    work page 2020

  34. [34]

    Sankhy¯ a: the indian journal of statistics, 401–406 (1946)

    Bhattacharyya, A.: On a measure of divergence between two m ultinomial populations. Sankhy¯ a: the indian journal of statistics, 401–406 (1946)

    work page 1946

  35. [35]

    IEEE transactions on communication technology 15(1), 52–60 (1967)

    Kailath, T.: The divergence and bhattacharyya distance measu res in signal selection. IEEE transactions on communication technology 15(1), 52–60 (1967)

    work page 1967

  36. [36]

    Bulletin of the Calcut ta Mathematical Society 35, 99–110 (1943)

    Bhattacharyya, A.: On a measure of divergence between two s tatistical pop- ulations defined by their probability distribution. Bulletin of the Calcut ta Mathematical Society 35, 99–110 (1943)

    work page 1943

  37. [37]

    Resonance 4(6), 20–26 (1999)

    McLachlan, G.J.: Mahalanobis distance. Resonance 4(6), 20–26 (1999)

    work page 1999

  38. [38]

    Multi-View Multiple Clustering

    Yao, S., Yu, G., Wang, J., Domeniconi, C., Zhang, X.: Multi-view multip le clustering. arXiv preprint arXiv:1905.05053 (2019)

    work page internal anchor Pith review Pith/arXiv arXiv 1905

  39. [39]

    Journal f¨ ur die reine und angewandte Mathematik 1909(136), 210–271 (1909)

    Hellinger, E.: Neue begr¨ undung der theorie quadratischer for men von unendlichvielen ver¨ anderlichen. Journal f¨ ur die reine und angewandte Mathematik 1909(136), 210–271 (1909)

    work page 1909

  40. [40]

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

    Lee, G.Y., Dam, T., Poenar, D.P., Duong, V.N., Ferdaus, M.M.: Hela-v fa: A hellinger distance-attention-based feature aggregation networ k for few-shot clas- sification. In: Proceedings of the IEEE/CVF Winter Conference on Applications of Computer Vision, pp. 2173–2183 (2024)

    work page 2024

  41. [41]

    arXiv preprint arXiv:2 509.11220 (2025)

    Lee, G.Y., Dam, T., Ferdaus, M.M., Poenar, D.P., Duong, V.N.: Anrot -helanet: 43 Adverserially and naturally robust attention-based aggregation n etwork via the hellinger distance for few-shot classification. arXiv preprint arXiv:2 509.11220 (2025)

    work page 2025

  42. [42]

    Journal of Computation al Science 51, 101314 (2021)

    Grzyb, J., Klikowski, J., Wo´ zniak, M.: Hellinger distance weighted e nsemble for imbalanced data stream classification. Journal of Computation al Science 51, 101314 (2021)

    work page 2021

  43. [43]

    In: 2017 7th International Confere nce on Com- munication Systems and Network Technologies (CSNT), pp

    Kumari, A., Thakar, U.: Hellinger distance based oversampling met hod to solve multi-class imbalance problem. In: 2017 7th International Confere nce on Com- munication Systems and Network Technologies (CSNT), pp. 137–14 1 (2017). IEEE

    work page 2017

  44. [44]

    arXiv preprint arXiv:2208.10559 (2022)

    Singh, A., Jamali-Rad, H.: Transductive decoupled variational inf erence for few- shot classification. arXiv preprint arXiv:2208.10559 (2022)

    work page arXiv 2022

  45. [45]

    In: Proceedings of the I EEE/CVF Conference on Computer Vision and Pattern Recognition, pp

    Wu, H., Qu, Y., Lin, S., Zhou, J., Qiao, R., Zhang, Z., Xie, Y., Ma, L.: Co ntrastive learning for compact single image dehazing. In: Proceedings of the I EEE/CVF Conference on Computer Vision and Pattern Recognition, pp. 1055 1–10560 (2021)

    work page 2021

  46. [46]

    arXiv preprint arXiv:2410.14595 (20 24)

    Lee, G.Y., Dam, T., Ferdaus, M.M., Poenar, D.P., Duong, V.: Draco- dehazenet: An efficient image dehazing network combining detail recovery and a n ovel contrastive learning paradigm. arXiv preprint arXiv:2410.14595 (20 24)

    work page arXiv

  47. [47]

    Advances in neural information proces sing systems 28 (2015)

    Kingma, D.P., Salimans, T., Welling, M.: Variational dropout and the lo cal reparameterization trick. Advances in neural information proces sing systems 28 (2015)

    work page 2015

  48. [48]

    In: Internation al Conference on Machine Learning, pp

    Chen, T., Kornblith, S., Norouzi, M., Hinton, G.: A simple framework for contrastive learning of visual representations. In: Internation al Conference on Machine Learning, pp. 1597–1607 (2020). PMLR

    work page 2020

  49. [49]

    Meta-learning with differentiable closed-form solvers

    Bertinetto, L., Henriques, J.F., Torr, P.H., Vedaldi, A.: Meta-lear ning with differentiable closed-form solvers. arXiv preprint arXiv:1805.08136 (2018)

    work page internal anchor Pith review Pith/arXiv arXiv 2018

  50. [50]

    arXiv preprint arXiv:2203.04291 (2022)

    Parnami, A., Lee, M.: Learning from few examples: A summary of a pproaches to few-shot learning. arXiv preprint arXiv:2203.04291 (2022)

    work page arXiv 2022

  51. [51]

    In: Proceed ings of the IEEE Conference on Computer Vision and Pattern Recognition, pp

    Sung, F., Yang, Y., Zhang, L., Xiang, T., Torr, P.H., Hospedales, T .M.: Learning to compare: Relation network for few-shot learning. In: Proceed ings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 1199 –1208 (2018)

    work page 2018

  52. [52]

    arxiv 2020

    Boudiaf, M., Ziko, I., Rony, J., Dolz, J., Piantanida, P., Ayed, I.: Tr ansduc- tive information maximization for few-shot learning. arxiv 2020. arX iv preprint arXiv:2008.11297 44

    work page arXiv 2020

  53. [53]

    arXiv preprint arXiv:1909.02729 (2019)

    Dhillon, G.S., Chaudhari, P., Ravichandran, A., Soatto, S.: A baselin e for few-shot image classification. arXiv preprint arXiv:1909.02729 (2019)

    work page arXiv 1909

  54. [54]

    In: Com- puter Vision–ECCV 2020: 16th European Conference, Glasgow, UK , August 23–28, 2020, Proceedings, Part I 16, pp

    Liu, J., Song, L., Qin, Y.: Prototype rectification for few-shot le arning. In: Com- puter Vision–ECCV 2020: 16th European Conference, Glasgow, UK , August 23–28, 2020, Proceedings, Part I 16, pp. 741–756 (2020). Sprin ger

    work page 2020

  55. [55]

    In: Proce edings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp

    Ding, Z., Xu, Y., Xu, W., Parmar, G., Yang, Y., Welling, M., Tu, Z.: Guide d variational autoencoder for disentanglement learning. In: Proce edings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , pp. 7920–7929 (2020)

    work page 2020

  56. [56]

    IEEE Transactions on Geoscienc e and Remote Sensing 57(2), 1155–1167 (2018)

    Wang, Q., Liu, S., Chanussot, J., Li, X.: Scene classification with re current atten- tion of vhr remote sensing images. IEEE Transactions on Geoscienc e and Remote Sensing 57(2), 1155–1167 (2018)

    work page 2018

  57. [57]

    ISPRS Journal of Photogrammetry and Remo te Sensing 209, 368–382 (2024)

    Qiu, C., Zhang, X., Tong, X., Guan, N., Yi, X., Yang, K., Zhu, J., Yu, A .: Few- shot remote sensing image scene classification: Recent advances, new baselines, and future trends. ISPRS Journal of Photogrammetry and Remo te Sensing 209, 368–382 (2024)

    work page 2024

  58. [58]

    IEEE Transactions on Geoscience and Remote Sensing 60, 1–14 (2021)

    Li, X., Deng, J., Fang, Y.: Few-shot object detection on remote sensing images. IEEE Transactions on Geoscience and Remote Sensing 60, 1–14 (2021)

    work page 2021

  59. [59]

    IEEE Tr ansactions on Geoscience and Remote Sensing (2023)

    Lang, C., Cheng, G., Tu, B., Han, J.: Global rectification and deco upled registra- tion for few-shot segmentation in remote sensing imagery. IEEE Tr ansactions on Geoscience and Remote Sensing (2023)

    work page 2023

  60. [60]

    IEEE Transactions on Pattern Analysis and Machine Intellige nce 45(4), 4650–4666 (2022)

    Cheng, G., Lang, C., Han, J.: Holistic prototype activation for fe w-shot segmen- tation. IEEE Transactions on Pattern Analysis and Machine Intellige nce 45(4), 4650–4666 (2022)

    work page 2022

  61. [61]

    Re mote Sensing 14(1), 111 (2021)

    Huang, W., Yuan, Z., Yang, A., Tang, C., Luo, X.: Tae-net: task- adaptive embed- ding network for few-shot remote sensing scene classification. Re mote Sensing 14(1), 111 (2021)

    work page 2021

  62. [62]

    Remote Sensing 13(13), 2532 (2021)

    Kim, J., Chi, M.: Saffnet: Self-attention-based feature fusion n etwork for remote sensing few-shot scene classification. Remote Sensing 13(13), 2532 (2021)

    work page 2021

  63. [63]

    IEEE Access 9, 19891–19901 (2020)

    Yuan, Z., Huang, W., Li, L., Luo, X.: Few-shot scene classification with multi- attention deepemd network in remote sensing. IEEE Access 9, 19891–19901 (2020)

    work page 2020

  64. [64]

    Wang, J., Shang, X., Yang, B., Ren, H., Jiang, H., Yao, H., Li, W.: Few -shot remote sensing image scene classification based on variational meta -learning (2025) 45

    work page 2025

  65. [65]

    IE EE Transactions on Geoscience and Remote Sensing (2024)

    Guo, Y., Fan, B., Feng, Y., Jia, X., He, M.: Distribution-aware and c lass-adaptive aggregation for few-shot hyperspectral image classification. IE EE Transactions on Geoscience and Remote Sensing (2024)

    work page 2024

  66. [66]

    Auto-Encoding Variational Bayes

    Kingma, D.P., Welling, M.: Auto-encoding variational bayes. arXiv p reprint arXiv:1312.6114 (2013)

    work page internal anchor Pith review Pith/arXiv arXiv 2013

  67. [67]

    In: Advances in Neural Information P rocessing Systems (2022)

    Roy, A., Shah, A., Shah, K., Dhar, P., Cherian, A., Chellappa, R.: Fe lmi: Few shot learning with hard mixup. In: Advances in Neural Information P rocessing Systems (2022)

    work page 2022

  68. [68]

    In: Proceedings of the IEEE/CVF Conference on Comput er Vision and Pattern Recognition, pp

    Sun, Q., Liu, Y., Chua, T.-S., Schiele, B.: Meta-transfer learning f or few-shot learning. In: Proceedings of the IEEE/CVF Conference on Comput er Vision and Pattern Recognition, pp. 403–412 (2019)

    work page 2019

  69. [69]

    Expert Systems with Applicatio ns 249, 123328 (2024)

    Sun, Z., Ying, W., Zhang, W., Gong, S.: Undersampling method base d on minor- ity class density for imbalanced data. Expert Systems with Applicatio ns 249, 123328 (2024)

    work page 2024

  70. [70]

    Advances in neural informatio n processing systems 35, 35504–35518 (2022)

    Park, C., Yun, S., Chun, S.: A unified analysis of mixed sample data a ugmen- tation: A loss function perspective. Advances in neural informatio n processing systems 35, 35504–35518 (2022)

    work page 2022

  71. [71]

    In: Proceedings of the IEEE/CVF Conferenc e on Computer Vision and Pattern Recognition, pp

    Lee, K., Maji, S., Ravichandran, A., Soatto, S.: Meta-learning wit h differentiable convex optimization. In: Proceedings of the IEEE/CVF Conferenc e on Computer Vision and Pattern Recognition, pp. 10657–10665 (2019)

    work page 2019

  72. [72]

    In: 2021 18th International Computer Confe rence on Wavelet Active Media Technology and Information Processing (ICCW AMTIP) , pp

    Tao, G., Weichao, L., Yanmin, H., Yu, L.: Graph-based prototypic al network for few-shot learning. In: 2021 18th International Computer Confe rence on Wavelet Active Media Technology and Information Processing (ICCW AMTIP) , pp. 234– 237 (2021). IEEE

    work page 2021

  73. [73]

    266–282 (2020)

    Tian, Y., Wang, Y., Krishnan, D., Tenenbaum, J.B., Isola, P.: Rethin king few-shot image classification: a good embedding is all you need? In: Computer V ision– ECCV 2020: 16th European Conference, Glasgow, UK, August 23– 28, 2020, Proceedings, Part XIV 16, pp. 266–282 (2020). Springer

    work page 2020

  74. [74]

    In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Reco gnition, pp

    Rizve, M.N., Khan, S., Khan, F.S., Shah, M.: Exploring complementar y strengths of invariant and equivariant representations for few-shot learnin g. In: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Reco gnition, pp. 10836–10846 (2021)

    work page 2021

  75. [75]

    In: P roceedings of the IEEE/CVF International Conference on Computer Vision, pp

    Ma, J., Xie, H., Han, G., Chang, S.-F., Galstyan, A., Abd-Almageed, W.: Partner-assisted learning for few-shot image classification. In: P roceedings of the IEEE/CVF International Conference on Computer Vision, pp. 10573–10582 (2021) 46

    work page 2021

  76. [76]

    arXiv preprint arXiv:200 6.09785 (2020)

    Rajasegaran, J., Khan, S., Hayat, M., Khan, F.S., Shah, M.: Self- supervised knowledge distillation for few-shot learning. arXiv preprint arXiv:200 6.09785 (2020)

    work page 2020

  77. [77]

    In: Pro- ceedings of the AAAI Conference on Artificial Intelligence, vol

    Jian, Y., Torresani, L.: Label hallucination for few-shot classific ation. In: Pro- ceedings of the AAAI Conference on Artificial Intelligence, vol. 36, pp. 7005–7014 (2022)

    work page 2022

  78. [78]

    I n: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Reco gnition, pp

    Zhang, C., Cai, Y., Lin, G., Shen, C.: Deepemd: Few-shot image clas sification with differentiable earth mover’s distance and structured classifiers. I n: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Reco gnition, pp. 12203–12213 (2020)

    work page 2020

  79. [79]

    In: Computer Vision–ECCV 2020: 16th Europea n Confer- ence, Glasgow, UK, August 23–28, 2020, Proceedings, Part V 16, pp

    Afrasiyabi, A., Lalonde, J.-F., Gagn´ e, C.: Associative alignment f or few-shot image classification. In: Computer Vision–ECCV 2020: 16th Europea n Confer- ence, Glasgow, UK, August 23–28, 2020, Proceedings, Part V 16, pp. 18–35 (2020). Springer

    work page 2020

  80. [80]

    In: Uncertaint y in Artificial Intelligence, pp

    Gao, Y., Fei, N., Liu, G., Lu, Z., Xiang, T.: Contrastive prototype le arning with augmented embeddings for few-shot learning. In: Uncertaint y in Artificial Intelligence, pp. 140–150 (2021). PMLR

    work page 2021

Showing first 80 references.