Research on Few-Shot Learning Based on Embedding Learning

doi:10.3778/j.issn.1002-8331.2108-0213

Abstract

Abstract: In order to solve the huge challenge of machine learning in the case of small sample size, researchers put forward the concept of few-shot learning. Among the existing research work, the embedding learning has achieved good results and aroused a lot of attention. Based on this background, the method is classified into independent embedding model and hybrid embedding model according to the way of combining task characteristics when training embedding function. On the basis of the classification, the existing methods of embedding learning are studied. In addition, the existing standard data sets of few-shot learning are summarized, the performance of each class of embedding learning methods are illustrated, and the factors affecting the learning performance of few-shot learning are analyzed. Finally, the current challenges of feature embedding learning are discussed, and the directions of future research are also prospected.

Key words: machine learning, few-shot learning, feature embedding function, embedding learning

摘要： 为了解决机器学习在样本量较少的情况下所面临的巨大挑战，研究人员提出了小样本学习的概念。在现有的小样本学习研究工作中，嵌入学习方法取得了不错的效果，引发了大量关注。根据训练特征嵌入函数时结合任务特征信息的方式，将嵌入学习方法划分为单一嵌入模型和混合嵌入模型两大类。依据划分的类别，对现有的嵌入学习方法的研究工作展开进行研究。汇总了现有的小样本标准数据集，阐述了每一类嵌入学习方法的表现，分析了影响小样本学习性能的因素。讨论嵌入学习方法目前面临的挑战，并展望未来的研究方向。

关键词: 机器学习, 小样本学习, 特征嵌入函数, 嵌入学习方法

HUANG Yanqian, CHI Dongxiang, XU Lingling. Research on Few-Shot Learning Based on Embedding Learning[J]. Computer Engineering and Applications, 2022, 58(3): 34-49.

黄彦乾, 迟冬祥, 徐玲玲. 面向小样本学习的嵌入学习方法研究综述[J]. 计算机工程与应用, 2022, 58(3): 34-49.

References

[1] HAN A T，RAMSUNDAR B，PAPPU A S，et al.Low data drug discovery with one-shot learning[J].ACS Central Science，2016，3（4）.
[2] MERAT F.Introduction to robotics：mechanics and control[J].IEEE Journal on Robotics and Automation，1987，3（2）：166.
[3] KAISER ?，NACHUM O，ROY A，et al.Learning to remember rare events[J].arXiv：1703.03129，2017.
[4] MITCHELL T M，THRUN S B.Explanation-based neural network learning for robot control[J].Advances in Neural Information Processing Systems，1993：287.
[5] LI Z，ZHOU F，CHEN F，et al.Meta-SGD：learning to learn quickly for few-shot learning[J].arXiv：1707.09835，2017.
[6] 田霞.基于元学习的少样本图像分类方法研究[D].成都：电子科技大学，2019.
TIAN X.A research of few-shot classificasion based on meta-learning[D].Chengdu：University of Electronic Science and Technology of China，2019.
[7] MITCHELL T.Machine learning[J].Annual Review of Computer Science，2003，4（1）：417-433.
[8] SHU J，XU Z，MENG D.Small sample learning in big data era[J].arXiv：1808.04572，2018.
[9] FINN C，ABBEEL P，LEVINE S.Model-agnostic meta learning for fast adaptation of deep networks[C]//Proceedings of International Conference on Machine Learning，2017：1126-1135.
[10] WANG Y，YAO Q，KWOK J T，et al.Generalizing from a few examples：a survey on few-shot learning[J].ACM Computing Surveys，2020，53（3）：1-34.
[11] LI F F，FERGUS R，PERONA P.One-shot learning of object categories[J].IEEE Transactions on Pattern Analysis and Machine Intelligence，2006，28（4）：594-611.
[12] FINK M.Object classification from a single example utilizing class relevance metrics[J].Advances in Neural Information Processing Systems，2005，17：449-456.
[13] HADDAD M，SANDERS D，LANGNER M，et al.One shot learning approach to identify drivers[C]//Proceedings of SAI Intelligent Systems Conference.Cham：Springer，2021：622-629.
[14] 刘颖，雷研博，范九伦，等.基于小样本学习的图像分类技术综述[J].自动化学报，2021，47（2）：297-315.
LIU Y，LEI Y B，FAN J L，et al.Survey on image classification technology based on ssmall sample learning[J].Acta Automatica Sinica，2021，47（2）：297-315.
[15] 任义丽，罗路.卷积神经网络过拟合问题研究[J].信息系统工程，2019（5）：140.
REN Y L，LUO L.Research on over fitting of convolutional neural network[J].China CID News，Information System Engineering，2019（5）：140.
[16] 聂金龙.基于度量学习的小样本学习研究[D].大连：大连理工大学，2020.
NIE J L.Research on few shot learning based on metric learning[D].Dalian：Dalian University of Technology，2020.
[17] MAHADEVAN S，TADEPALLI P.Quantifying prior determination knowledge using the pac learning model[J].Machine Learning，1994，17（1）：69-105.
[18] BOTTOU L，BOUSQUET O.The tradeoffs of large-scale learning[C]//Advances in Neural Information Processing System，2007.
[19] SANTORO A，BARTUNOV S，BOTVINICK M，et al.Meta-learning with memory-augmented neural networks[C]//International Conference on Machine Learning，2016：1842-1850.
[20] CARUANA R A.Multitask learning[M].[S.l.]：Kluwer Academic Publishers，1998.
[21] ARNOLD S M R，SHA F.Embedding adaptation is still needed for few-shot learning[J].arXiv：2104.07255，2021.
[22] FAN Y，LI Y，ZHU A.A few-shot learning algorithm based on attention adaptive mechanism[J].Journal of Physics：Conference Series，2021，1966（1）：012011.
[23] SEO J W，JUNG H G，LEE S W.Self-augmentation：Generalizing deep networks to unseen classes for few-shot learning[J].Neural Networks，2021（12）.
[24] BOTTOU L，CURTIS F E，NOCEDAL J.Optimization methods for large-scale machine learning[J].Siam Review，2018，60（2）：223-311.
[25] LI C，LI S，ZHANG A，et al.Meta-learning for few-shot bearing fault diagnosis under complex working conditions[J].Neurocomputing，2021，439（2）.
[26] LIU J Z.Small sample bark image recognition method based on convolutional neural network[J].Journal of Northwest Forestry University，2019，34（4）：230-235.
[27] HS A，MTT B，EZ A，et al.L2AE-D：learning to aggregate embeddings for few-shot learning with meta-level dropout-science direct[J].Neurocomputing，2021，442（4）：200-208.
[28] JIANG R，ZHANG J，YAN R，et al.Few-shot learning in spiking neural networks by multi-timescale optimization[J].Neural Computation，2021：1-34.
[29] BROCK A，LIM T，RITCHIE J M，et al.SMASH：one-shot model architecture search through hyperNetworks[J].arXiv：1708.05344，2017.
[30] LI Y，SHAO Z，HUANG X，et al.Meta-FSEO：a meta-learning fast adaptation with self-supervised embedding optimization for few-shot remote sensing scene classification[J].Remote Sensing，2021，13（14）：2776.
[31] HOSPEDALES T，ANTONIOU A，MICAELLI P，et al.Meta-learning in neural networks：a survey[J].arXiv：2004.05439，2020.
[32] 赵凯琳，靳小龙，王元卓.小样本学习研究综述[J].软件学报，2021，32（2）：349-369.
ZHAO K L，JIN X L，WANG Y Z.Survey on few-shot learning[J].Journal of Software，2021，32（2）：349-369.
[33] YE H J，HU H，ZHAN D C.Learning adaptive classifiers synthesis for generalized few-shot learning[J].International Journal of Computer Vision，2021，129（6）：1930-1953.
[34] YAN W，YAP J，MORI G.Multi-task transfer methods to improve one-shot learning for multimedia event detection[C]//British Machine Vision Conference，2015.
[35] 祝钧桃，姚光乐，张葛祥，等.深度神经网络的小样本学习综述[J].计算机工程与应用，2021，57（7）：22-33.
ZHU J T，YAO G L，ZHANG G X，et al.Survey of few shot learning of deep neural network[J].Computer Engineering and Applications，2021，57（7）：22-33.
[36] BROMLEY J，BENTZ J W，BOTTOU L，et al.Signature verification using a “siamese” time delay neural network[J].International Journal of Pattern Recognition and Artificial Intelligence，1993，7（4）：669-688.
[37] KOCH G，ZEMEL R，SALAKHUTDINOV R.Siamese neural networks for one-shot image recognition[C]//International Conference on Machine Learning，2015.
[38] SAMEER V U，NASKAR R.Deep siamese network for limited labels classification in source camera identification[J].Multimedia Tools and Applications，2020，79（3）：28079-28104.
[39] ZHAO L C，SHANG Z W，ZHAO L，et al.Siamese dense neural network for software defect prediction with small data[J].IEEE Access，2019，7：7663-7677.
[40] VINYALS O，BLUNDELL C，LILLICRAP T，et al.Matching networks for one shot learning[J].Advances in Neural Information Processing Systems，2016，29：3630-3638.
[41] CAI Q，PAN Y W，YAO T，et al.Memory matching networks for one-shot image recognition[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition，2018.
[42] HOCHREITER S，SCHMIDHUBER J.Long short-term memory[J].Neural Computation，1997，9（8）：1735-1780.
[43] LI H，DONG W，MEI X，et al.Lgm-net：learning to generate matching networks for few-shot learning[C]//International Conference on Machine Learning，2019：3825-3834.
[44] HOU R，CHANG H，MA B，et al.Cross attention network for few-shot classification[J].arXiv：1910.07677，2019.
[45] CHEN Y，DAI X，LIU M，et al.Dynamic convolution：attention over convolution kernels[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition，2020：11030-11039.
[46] JIANG L B，ZHOU X L，JIANG F W，et al.One-shot learning based on improved matching network[J].Systems Engineering and Electronics，2019，41（6）：1210-1217.
[47] SLEDGE I J，TOOLE C D，MAESTRI J A，et al.External-memory networks for low-shot learning of targets in forward-looking-sonar imagery[J].arXiv：2107.10504，2021.
[48] BACHMAN P，SORDONI A，TRISCHLER A.Learning algorithms for active learning[C]//International Conference on Machine Learning，2017：301-310.
[49] COHN D A，GHAHRAMANI Z，JORDAN M I.Active learning with statistical models[J].Journal of Artificial Intelligence Research，1996，4：129-145.
[50] WANG P，LIU L，SHEN C，et al.Multi-attention network for one shot learning[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2017：2721-2729.
[51] SNELL J，SWERSKY K，ZEMEL R S.Prototypical networks for few-shot learning[J].arXiv：1703.05175，2017.
[52] WANG Y X，GIRSHICK R，HEBERT M，et al.Low-shot learning from imaginary data[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2018：7278-7286.
[53] TANG H，LI Y，HAN X，et al.A spatial-spectral prototypical network for hyperspectral remote sensing image[J].IEEE Geoscience and Remote Sensing Letters，2019，17（1）：167-171.
[54] GAO T，HAN X，LIU Z，et al.Hybrid attention-based prototypical networks for noisy few-shot relation classification[C]//Proceedings of the AAAI Conference on Artificial Intelligence，2019：6407-6414.
[55] REN M，TRIANTAFILLOU E，RAVI S，et al.Meta-learning for semi-supervised few-shot classification[J].arXiv：1803.00676，2018.
[56] LIU J W，LIU Y，LUO X L.Semi-supervised learning methods[J].Chinese Journal of Computers，2015，38（8）：1592-1617.
[57] SUN S，SUN Q，ZHOU K，et al.Hierarchical attention prototypical networks for few-shot text classification[C]//Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing and the 9th International Joint Conference on Natural Language Processing（EMNLP-IJCNLP），2019：476-485.
[58] SETH H，KUMAR P，SRIVASTAVA M M.Prototypical metric transfer learning for continuous speech keyword spotting with limited training data[C]//International Workshop on Soft Computing Models in Industrial and Environmental Applications.Cham：Springer，2019：273-280.
[59] PAN S J，QIANG Y.A survey on transfer learning[J].IEEE Transactions on Knowledge and Data Engineering，2010，22（10）：1345-1359.
[60] SUNG F，YANG Y，ZHANG L，et al.Learning to compare：relation network for few-shot learning[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition，2018：1199-1208.
[61] ZHANG X，SUNG F.RelationNet2：deep comparison columns for few-shot learning[J].arXiv：1811.07100，2018.
[62] HE J，HONG R，LIU X，et al.Memory-augmented relation network for few-shot learning[C]//Proceedings of the 28th ACM International Conference on Multimedia，2020：1236-1244.
[63] SUKHBAATAR S，SZLAM A，WESTON J，et al.End-to-end memory networks[J].arXiv：1503.08895，2015.
[64] INOUE H.Data augmentation by pairing samples for images classification[J].arXiv：1801.02929，2018.
[65] SCARSELLI F，GORI M，TSOI A C，et al.The graph neural network model[J].IEEE Transactions on Neural Networks，2008，20（1）：61-80.
[66] ZHOU J，CUI G，HU S，et al.Graph neural networks：a review of methods and applications[J].arXiv：1812.08434，2018.
[67] DEFFERRARD M，BRESSON X，VANDERGHEYNST P.Convolutional neural networks on graphs with fast localized spectral filtering[J].Advances in Neural Information Processing Systems，2016，29：3844-3852.
[68] BRUNA J，ZAREMBA W，SZLAM A，et al.Spectral networks and locally connected networks on graphs[J].arXiv：1312.6203，2013.
[69] LIU Y，LEI Y，RASHID S F.Graph convolution network with node feature optimization using cross attention for few-shot learning[C]//Proceedings of the 2nd ACM International Conference on Multimedia in Asia，2021：1-7.
[70] GARCIA V，BRUNA J.Few-shot learning with graph neural networks[J].arXiv：1711.04043，2017.
[71] GIDARIS S，KOMODAKIS N.Generating classification weights with gnn denoising autoencoders for few-shot learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition，2019：21-30.
[72] YAO H，ZHANG C，WEI Y，et al.Graph few-shot learning via knowledge transfer[J].arXiv：1910.03053，2019.
[73] LIU X，LIU P，ZONG L.Transductive prototypical network for few-shot classification[C]//2020 IEEE International Conference on Image Processing（ICIP），2020.
[74] LIU Y，LEE J，PARK M，et al.Learning to propagate labels：transductive propagation network for few-shot learning[J].arXiv：1805.10002，2018.
[75] KIM J，KIM T，KIM S，et al.Edge-labeling graph neural network for few-shot learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition，2019：11-20.
[76] TRIANTAFILLOU E，ZEMEL R，URTASUN R.Few-shot learning through an information retrieval lens[J].arXiv：1707.02610，2017.
[77] BERTINETTO L，HENRIQUES J F，VALMADRE J，et al.Learning feed-forward one-shot learners[C]//Advances in Neural Information Processing Systems，2016：523-531.
[78] BERTINETTO L，HENRIQUES J F，TORR P H S，et al.Meta-learning with differentiable closed-form solvers[J].arXiv：1805.08136，2018.
[79] ZHAO F，ZHAO J，YAN S，et al.Dynamic conditional networks for few-shot learning[C]//Proceedings of the European Conference on Computer Vision（ECCV），2018：19-35.
[80] RAKELLY K，SHELHAMER E，DARRELL T，et al.Conditional networks for few-shot semantic segmentation[C]//International Conference on Learning Representations，2018.
[81] GIDARIS S，KOMODAKIS N.Dynamic few-shot visual learning without forgetting[C]//2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition （CVPR），2018.
[82] ORESHKIN B N，RODRIGUEZ P，LACOSTE A.Tadam：task dependent adaptive metric for improved few-shot learning[J].arXiv：1805.10123，2018.
[83] BUHRMESTER M D，KWANG T，GOSLING S D.Amazon’s mechanical turk[J].Perspectives on Psychological Science，2011，6（1）：3-5.
[84] FORT S.Gaussian prototypical networks for few-shot learning on omniglot[J].arXiv：1708.02735，2017.
[85] MALALUR P，JAAKKOLA T.Alignment based matching networks for one-shot classification and open-set recognition[J].arXiv：1903.06538，2019.
[86] TORRALBA A，FERGUS R，FREEMAN W T.80 million tiny images：a large data set for nonparametric object and scene recognition[J].IEEE Transactions on Pattern Analysis & Machine Intelligence，2008，30（11）：1958-1970.
[87] BRUNKE L，GREEFF M，HALL A W，et al.Safe learning in robotics：from learning-based control to safe reinforcement learning[J].arXiv：2108.06266，2021.
[88] FENG R W，ZHENG X S，GAO T X，et al.Interactive few-shot learning：limited supervision，better medical image segmentation[J].IEEE Transactions on Medical Imaging，2021，40（10）：2575-2588.
[89] HAN X，ZHU H，YU P，et al.Fewrel：a large-scale supervised few-shot relation classification dataset with state-of-the-art evaluation[J].arXiv：1810.10147，2018.
[90] JIANG D，WANG R，YANG J，et al.Kernel multi-attention neural network for knowledge graph embedding[J].Knowledge-Based Systems，2021（7）：107188.