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.
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
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
Ye M, uhong Guo (2018) Deep triplet ranking networks for one-shot recognition. CoRR abs/1804.07275. http://arxiv.org/abs/1804.07275
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
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
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
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
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
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
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
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
Acknowledgements
This work was supported by the National Social Science Fund of China (No. 22ZDA121).
Author information
Authors and Affiliations
Corresponding author
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.
About this article
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
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s00521-022-08080-w