Skip to main content
SpringerLink
Log in

NC\(^2\)E: boosting few-shot learning with novel class center estimation

  • Original Article
  • Published:
Neural Computing and Applications Aims and scope Submit manuscript

Abstract

Accurate class distribution estimation is expected to solve the problem of the poor generalization ability that exists in few-shot learning models due to data shortages. However, the reliability of class distributions estimates based on limited samples and knowledge is questionable, especially for similar classes. We find that the distribution calibration method is inaccurate in estimating similar classes due to limited knowledge being reused through double-validation experiments. To address this issue, we propose a novel class center estimation (NC\(^2\)E) method, which consists of a two-stage center estimation (TCE) algorithm and a class centroid estimation (CCE) algorithm. The class centers estimated by TCE in two stages are closer to the truth, and its superiority is demonstrated by error theory. CCE searches for the centroid of the base class iteratively and is used as the basis for the novel class calibration. Sufficient simulation samples are generated based on the estimated class distribution to augment the training data. The experimental results show that, compared with the distribution calibration method, the proposed method achieves an approximately 1% performance improvement on the miniImageNet and CUB datasets; an approximately 1.45% performance improvement for similar class classification; and an approximately 6.06% performance improvement for non-similar class classification.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6

Similar content being viewed by others

Data availability statement

The public datasets which we use in our experiments is available https://www.image-net.org and http://www.vision.caltech.edu/datasets/.

References

  1. Andrychowicz M, Denil M, Colmenarejo SG, et al (2016) Learning to learn by gradient descent by gradient descent. In: Proceedings of the 30th international conference on neural information processing systems. Curran Associates Inc., NIPS’16, pp 3988–3996. https://doi.org/10.5555/3157382.3157543

  2. Chen WY, Liu YC, Kira Z, et al (2019) A closer look at few-shot classification. In: International conference on learning representations (ICLR). https://openreview.net/forum?id=HkxLXnAcFQ

  3. Chen Z, Fu Y, Zhang Y et al (2019) Multi-level semantic feature augmentation for one-shot learning. IEEE Trans Image Process (TIP) 28(9):4594–4605. https://doi.org/10.1109/TIP.2019.2910052

    Article  MathSciNet  MATH  Google Scholar 

  4. Das D, Lee CSG (2020) A two-stage approach to few-shot learning for image recognition. IEEE Trans Image Process (TIP) 29:3336–3350. https://doi.org/10.1109/TIP.2019.2959254

    Article  MATH  Google Scholar 

  5. Demšar J (2006) Statistical comparisons of classifiers over multiple data sets. J Mach Learn Res 7:1–30. https://doi.org/10.5555/1248547.1248548

    Article  MathSciNet  MATH  Google Scholar 

  6. Duchi J, Hazan E, Singer Y (2011) Adaptive subgradient methods for online learning and stochastic optimization. J Mach Learn Res 12:2121–2159. https://doi.org/10.5555/1953048.2021068

    Article  MathSciNet  MATH  Google Scholar 

  7. Finn C, Abbeel P, Levine S (2017) Model-agnostic meta-learning for fast adaptation of deep networks. In: International conference on machine learning (ICML), pp 1126–1135. https://proceedings.mlr.press/v70/finn17a.html

  8. Gidaris S, Komodakis N (2018) Dynamic few-shot visual learning without forgetting. In: 2018 IEEE/CVF conference on computer vision and pattern recognition (CVPR), pp 4367–4375. https://doi.org/10.1109/CVPR.2018.00459

  9. Goodfellow I, Pouget-Abadie J, Mirza M, et al (2014) Generative adversarial nets. In: Advances in neural information processing systems (NeurIPS). https://doi.org/10.5555/2969033.2969125

  10. Gou J, Song J, Du L et al (2021) Class mean-weighted discriminative collaborative representation for classification. Int J Intell Syst 36(7):3144–3173. https://doi.org/10.1002/int.22411

    Article  Google Scholar 

  11. Gou J, He X, Lu J et al (2022) A class-specific mean vector-based weighted competitive and collaborative representation method for classification. Neural Netw 150:12–27. https://doi.org/10.1016/j.neunet.2022.02.021

    Article  Google Scholar 

  12. Hu Y, Gripon V, Pateux S (2021) Leveraging the feature distribution in transfer-based few-shot learning. In: International conference on artificial neural networks (ICANN), pp 487–499. https://doi.org/10.1007/978-3-030-86340-1_39

  13. Jiang H, Wang R, Shan S, et al (2018) Learning class prototypes via structure alignment for zero-shot recognition. In: Proceedings of the European conference on computer vision (ECCV), pp 118–134. https://doi.org/10.1007/978-3-030-01249-6_8

  14. Jiang L, Zhang L, Li C et al (2019) A correlation-based feature weighting filter for Naive Bayes. IEEE Trans Knowl Data Eng 31(2):201–213. https://doi.org/10.1109/TKDE.2018.2836440

    Article  Google Scholar 

  15. Jiang S, Zhu Y, Liu C et al (2022) Dataset bias in few-shot image recognition. IEEE Trans Pattern Anal Mach Intell (TPAMI). https://doi.org/10.1109/TPAMI.2022.3153611

    Article  Google Scholar 

  16. Jung HG, Lee SW (2020) Few-shot learning with geometric constraints. IEEE Trans Neural Netw Learn Syst (TNNLS) 31(11):4660–4672. https://doi.org/10.1109/TNNLS.2019.2957187

    Article  Google Scholar 

  17. Li Z, Zhou F, Chen F, et al (2017) Meta-sgd: learning to learn quickly for few-shot learning. arXiv:1707.09835, https://doi.org/10.48550/arXiv.1707.09835

  18. Liu B, Cao Y, Lin Y, et al (2020) Negative margin matters: Understanding margin in few-shot classification. In: European conference on computer vision (ECCV), pp 438–455. https://doi.org/10.1007/978-3-030-58548-8_26

  19. Liu Y, Schiele B, Sun Q (2020) An ensemble of epoch-wise empirical bayes for few-shot learning. In: European conference on computer vision (ECCV), pp 404–421. https://doi.org/10.1007/978-3-030-58517-4_24

  20. Liu Z, Miao Z, Zhan X, et al (2019) Large-scale long-tailed recognition in an open world. In: 2019 IEEE/CVF conference on computer vision and pattern recognition (CVPR), pp 2532–2541. https://doi.org/10.1109/CVPR.2019.00264

  21. Mangla P, Kumari N, Sinha A, et al (2020) Charting the right manifold: manifold mixup for few-shot learning. In: Proceedings of the IEEE/CVF winter conference on applications of computer vision (WACV), pp 2218–2227. https://doi.org/10.1109/WACV45572.2020.9093338

  22. Mygdalis V, Pitas I (2022) Hyperspherical class prototypes for adversarial robustness. Pattern Recogn 125(108):527. https://doi.org/10.1016/j.patcog.2022.108527

    Article  Google Scholar 

  23. Oreshkin BN, Rodriguez P, Lacoste A (2018) Tadam: Task dependent adaptive metric for improved few-shot learning. In: Proceedings of the 32nd international conference on neural information processing systems. Curran Associates Inc., Red Hook, NIPS’18, pp 719–729. https://doi.org/10.5555/3326943.3327010

  24. Ouyang C, Biffi C, Chen C et al (2022) Self-supervised learning for few-shot medical image segmentation. IEEE Trans Med Imaging (TMI). https://doi.org/10.1109/TMI.2022.3150682

    Article  Google Scholar 

  25. Park SJ, Han S, Baek JW, et al (2020) Meta variance transfer: learning to augment from the others. In: International conference on machine learning (ICML), pp 7510–7520. https://doi.org/10.5555/3524938.3525634

  26. Rusu AA, Rao D, Sygnowski J, et al (2019) Meta-learning with latent embedding optimization. In: International conference on learning representations (ICLR). https://openreview.net/forum?id=BJgklhAcK7

  27. Schwartz E, Karlinsky L, Shtok J, et al (2018) Delta-encoder: an effective sample synthesis method for few-shot object recognition. In: Advances in neural information processing systems (NeurIPS). https://proceedings.neurips.cc/paper/2018/file/1714726c817af50457d810aae9d27a2e-Paper.pdf

  28. Shao S, Wang P, Yan R (2019) Generative adversarial networks for data augmentation in machine fault diagnosis. Comput Ind 106:85–93. https://doi.org/10.1016/j.compind.2019.01.001

    Article  Google Scholar 

  29. Simon C, Koniusz P, Nock R, et al (2020) Adaptive subspaces for few-shot learning. In: 2020 IEEE/CVF conference on computer vision and pattern recognition (CVPR), pp 4136–4145. https://doi.org/10.1109/CVPR42600.2020.00419

  30. Snell J, Swersky K, Zemel R (2017) Prototypical networks for few-shot learning. In: Advances in neural information processing systems (NeurIPS). https://proceedings.neurips.cc/paper/2017/file/cb8da6767461f2812ae4290eac7cbc42-Paper.pdf

  31. Su J, Shen H, Peng L et al (2021) Few-shot domain-adaptive anomaly detection for cross-site brain images. IEEE Trans Pattern Anal Mach Intell (TPAMI). https://doi.org/10.1109/TPAMI.2021.3125686

    Article  Google Scholar 

  32. Sun Q, Liu Y, Chua TS, et al (2019) Meta-transfer learning for few-shot learning. In: 2019 IEEE/CVF conference on computer vision and pattern recognition (CVPR), pp 403–412. https://doi.org/10.1109/CVPR.2019.00049

  33. Vinyals O, Blundell C, Lillicrap T, et al (2016) Matching networks for one shot learning. In: Advances in neural information processing systems (NeurIPS), pp 3630–3638. https://proceedings.neurips.cc/paper/2016/file/90e1357833654983612fb05e3ec9148c-Paper.pdf

  34. Wah C, Branson S, Welinder P, et al (2011) The caltech-ucsd birds-200-2011 dataset. Tech. Rep. CNS-TR-2011-001, California Institute of Technology

  35. Wang P, Cheng J, Hao F et al (2020) Embedded adaptive cross-modulation neural network for few-shot learning. Neural Comput Appl 32(10):5505–5515. https://doi.org/10.1007/s00521-019-04605-y

    Article  Google Scholar 

  36. Xue W, Wang W (2020) One-shot image classification by learning to restore prototypes. Proc AAAI Conf Artif Intell (AAAI) 34:6558–6565. https://doi.org/10.1609/aaai.v34i04.6130

    Article  MathSciNet  Google Scholar 

  37. Yang S, Liu L, Xu M (2021) Free lunch for few-shot learning: distribution calibration. In: International conference on learning representations (ICLR). https://openreview.net/forum?id=JWOiYxMG92s

  38. Ye M, uhong Guo (2018) Deep triplet ranking networks for one-shot recognition. CoRR abs/1804.07275. http://arxiv.org/abs/1804.07275

  39. Zagoruyko S, Komodakis N (2016) Wide residual networks. In: Proceedings of the British machine vision conference (BMVC), pp 87.1–87.12. https://doi.org/10.5244/C.30.87

  40. Zhang H, Cao Z, Yan Z, et al (2021) Sill-net: Feature augmentation with separated illumination representation. arXiv:2102.03539, https://doi.org/10.48550/arXiv.2102.03539

  41. Zhang J, Zhao C, Ni B, et al (2019) Variational few-shot learning. In: Proceedings of the IEEE/CVF international conference on computer vision (ICCV), pp 1685–1694. https://doi.org/10.1109/ICCV.2019.00177

  42. Zhang R, Che T, Ghahramani Z, et al (2018) Metagan: An adversarial approach to few-shot learning. In: Proceedings of the 32nd international conference on neural information processing systems. Curran Associates Inc., NIPS’18, pp 2371–2380. https://doi.org/10.5555/3327144.3327163

  43. Zhang R, Che T, Ghahramani Z, et al (2018) Metagan: an adversarial approach to few-shot learning. In: Advances in neural information processing systems (NeurIPS). https://proceedings.neurips.cc/paper/2018/file/4e4e53aa080247bc31d0eb4e7aeb07a0-Paper.pdf

  44. Zhang Z (2018) Improved adam optimizer for deep neural networks. In: 2018 IEEE/ACM 26th international symposium on quality of service (IWQoS), pp 1–2. https://doi.org/10.1109/IWQoS.2018.8624183

  45. Zhu X, Guo K, Fang H et al (2021) Cross view capture for stereo image super-resolution. IEEE Trans Multimedia 24:3074–3086. https://doi.org/10.1109/TMM.2021.3092571

    Article  Google Scholar 

  46. Zhu X, Guo K, Ren S et al (2022) Lightweight image super-resolution with expectation-maximization attention mechanism. IEEE Trans Circuits Syst Video Technol 32(3):1273–1284. https://doi.org/10.1109/TCSVT.2021.3078436

    Article  Google Scholar 

Download references

Acknowledgements

This work was supported by the National Social Science Fund of China (No. 22ZDA121).

Author information

Authors and Affiliations

  1. School of Information Engineering, Hunan University of Science and Engineering, Yongzhou, 425199, China

    Zheng Wu & Entao Luo

  2. School of Computer Science and Engineering, Central South University, Changsha, 410083, China

    Zheng Wu, Changchun Shen, Kehua Guo & Liwei Wang

Authors
  1. Zheng Wu
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Changchun Shen
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Kehua Guo
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Entao Luo
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Liwei Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Kehua Guo.

Ethics declarations

Conflict of interest

The authors declare that they have no known competing financial interests or personal relations that could have appeared to influence the work reported in this paper.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Springer Nature or its licensor (e.g. a society or other partner) holds exclusive rights to this article under a publishing agreement with the author(s) or other rightsholder(s); author self-archiving of the accepted manuscript version of this article is solely governed by the terms of such publishing agreement and applicable law.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Wu, Z., Shen, C., Guo, K. et al. NC\(^2\)E: boosting few-shot learning with novel class center estimation. Neural Comput & Applic 35, 7049–7062 (2023). https://doi.org/10.1007/s00521-022-08080-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00521-022-08080-w

Keywords

Search

Navigation