计算机科学 ›› 2023, Vol. 50 ›› Issue (6): 225-235.doi: 10.11896/jsjkx.220900197
蔡少填, 陈小军, 陈龙腾, 邱莉萍
CAI Shaotian, CHEN Xiaojun, CHEN Longteng, QIU Liping
摘要: 图像聚类是图像处理中一个重要且开放的问题。最近,一些方法利用联合对比学习的良好表征能力来进行端到端聚类学习,利用伪标签技术来生成高质量的伪标签以提升聚类模型的鲁棒性。伪标签方法通常需要设置一个较大的概率阈值,并对满足要求的样本生成one-hot的标签,同时利用生成的标签来更新模型。但是,这种简单的伪标签生成方法难以获得足够数量的高质量伪标签。为了解决以上问题,提出了一种基于分层伪标签的图像聚类方法,它旨在利用结构化信息与伪标签信息对分类模型进行训练和精炼。引入3个假设来指导聚类方法的设计,包括局部平滑假设、自训练假设及低密度分离假设。新方法包含两个阶段:1)基于流形的一致性学习,利用近邻一致性学习来初始化聚类模型;2)基于分层伪标签的模型精炼,基于第一阶段的结果生成伪标签,并利用其来提升聚类模型的鲁棒性。首先,将基于第一阶段的结果生成强伪标签数据集及弱伪标签数据集;然后,提出了基于标签传播及分层混合的伪标签提升技术来提升弱伪标签数据集的质量;最后,同时利用强伪标签数据集及弱伪标签数据集来提升分类模型的泛化能力。相较于最优结果,SPC算法在STL10和Cifar100-20基准数据集上,ACC平均结果分别提升了7.6%和5.0%。
中图分类号:
[1]XIE J,GIRSHICK R,FARHADI A.Unsupervised deep embedding for clustering analysis[C]//International Conference on Machine Learning.PMLR,2016:478-487. [2]YANG B,FU X,SIDIROPOULOS N D,et al.Towards k-means-friendly spaces:Simultaneous deep learning and clustering[C]//International Conference on Machine Learning.PMLR,2017:3861-3870. [3]TIAN K,ZHOU S,GUAN J.DEEPCLUSTER:A general clustering framework based on deep learning[C]//Joint European Conference on Machine Learning and Knowledge Discovery in Databases.Cham:Springer,2017:809-825. [4]AGGARWAL C C,WOLF J L,YU P S,et al.Fast algorithms for projected clustering[J].ACM SIGMoD Record,1999,28(2):61-72. [5]SHAHAM U,STANTON K,LI H,et al.Spectralnet:Spectralclustering using deep neural networks[J].arXiv:1801.01587,2018. [6]YANG J,PARIKH D,BATRA D.Joint unsupervised learning of deep representations and image clusters[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016:5147-5156. [7]CHEN T,KORNBLITH S,NOROUZI M,et al.A simpleframework for contrastive learning of visual representations[C]//International Conference on Machine Learning.PMLR,2020:1597-1607. [8]HE K,FAN H,WU Y,et al.Momentum contrast for unsupervised visual representation learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2020:9729-9738. [9]JI X,HENRIQUES J F,VEDALDI A.Invariant informationclustering for unsupervised image classification and segmentation[C]//Proceedings of the IEEE/CVF International Confe-rence on Computer Vision.2019:9865-9874. [10]LI Y,HU P,LIU Z,et al.Contrastive clustering[C]//Procee-dings of the AAAI Conference on Artificial Intelligence.2021:8547-8555. [11]DO K,TRAN T,VENKATESH S.Clustering by maximizingmutual information across views[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.2021:9928-9938. [12]DANG Z,DENG C,YANG X,et al.Nearest neighbor matching for deep clustering[C]//Proceedings of the IEEE/CVF Confe-rence on Computer Vision and Pattern Recognition.2021:13693-13702. [13]ZHANG R,ISOLA P,EFROS A A.Colorful image colorization[C]//European Conference on Computer Vision.Cham:Sprin-ger,2016:649-666. [14]GIDARIS S,SINGH P,KOMODAKIS N.Unsupervised representation learning by predicting image rotations[J].arXiv:1803.07728,2018. [15]PATHAK D,KRAHENBUHL P,DONAHUE J,et al.Context encoders:Feature learning by inpainting[C]//Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition.2016:2536-2544. [16]VAN GANSBEKE W,VANDENHENDE S,GEORGOULIS S,et al.Scan:Learning to classify images without labels[C]//European Conference on Computer Vision.Cham:Springer,2020:268-285. [17]JI P,ZHANG T,LI H,et al.Deep subspace clustering networks[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems.2017:23-32. [18]YAMAGUCHI M,IRIE G,KAWANISHI T,et al.Subspacestructure-aware spectral clustering for robust subspace clustering[C]//Proceedings of the IEEE/CVF International Confe-rence on Computer Vision.2019:9875-9884. [19]LAW M T,URTASUN R,ZEMEL R S.Deep spectral clustering learning[C]//International Conference on Machine Lear-ning.PMLR,2017:1985-1994. [20]YANG X,DENG C,ZHENG F,et al.Deep spectral clusteringusing dual autoencoder network[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2019:4066-4075. [21]ZHONG H,WU J,CHEN C,et al.Graph contrastive clustering[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.2021:9224-9233. [22]WU J,LONG K,WANG F,et al.Deep comprehensive correlation mining for image clustering[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision.2019:8150-8159. [23]CHANG J,WANG L,MENG G,et al.Deep adaptive imageclustering[C]//Proceedings of the IEEE International Confe-rence on Computer Vision.2017:5879-5887. [24]MAHON L,LUKASIEWICZT.Selective Pseudo-Label Clustering[C]//German Conference on Artificial Intelligence (Künstliche Intelligenz).Cham:Springer,2021:158-178. [25]GUPTA D,RAMJEE R,KWATRA N,et al.Unsupervised clustering using pseudo-semi-supervised learning[C]//International Conference on Learning Representations.2020. [26]PARK S,HAN S,KIM S,et al.Improving unsupervised image clustering with robust learning[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2021:12278-12287. [27]LOWE D G.Object recognition from local scale-invariant fea-tures[C]//Proceedings of the seventh IEEE International Conference on Computer Vision.1999:1150-1157. [28]DALAL N,TRIGGS B.Histograms of oriented gradients forhuman detection[C]//2005 IEEEComputer Society Conference on Computer Vision and Pattern Recognition (CVPR'05).2005:886-893. [29]VINCENT P,LAROCHELLE H,BENGIO Y,et al.Extracting and composing robust features with denoising autoencoders[C]//Proceedings of the 25th International Conference on Machine Learning.2008:1096-1103. [30]DONAHUE J,SIMONYAN K.Large scale adversarial representation learning[C]//Proceedings of the 33rd International Conference on Neural Information Processing Systems.2019:10542-10552. [31]NOROOZI M,PIRSIAVASH H,FAVARO P.Representationlearning by learning to count[C]//Proceedings of the IEEE International Conference on Computer Vision.2017:5898-5906. [32]BACHMAN P,HJELM R D,BUCHWALTER W.Learningrepresentations by maximizing mutual information across views[C]//Proceedings of the 33rd International Conference on Neural Information Processing Systems.2019:15535-15545. [33]HADSELL R,CHOPRA S,LECUN Y.Dimensionality reduction by learning an invariant mapping[C]//2006 IEEE Compu-ter Society Conference on Computer Vision and Pattern Recognition (CVPR'06).IEEE,2006:1735-1742. [34]GRILL J B,STRUB F,ALTCHÉ F,et al.Bootstrap your own latent-a new approach to self-supervised learning[J].Advances in Neural Information Processing Systems,2020,33:21271-21284. [35]LI J,ZHOU P,XIONG C,et al.Prototypical Contrastive Lear-ning of Unsupervised Representations[C]//ICLR.2021. [36]GUO Y,XU M,LI J,et al.HCSC:Hierarchical Contrastive Selective Coding[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2022:9706-9715. [37]SAJJADI M,JAVANMARDI M,TASDIZEN T.Regularization with stochastic transformations and perturbations for deep semi-supervised learning[C]//Proceedings of the 30th International Conference on Neural Information Processing Systems.2016:1171-1179. [38]OORD A,LI Y,VINYALS O.Representation learning with contrastive predictive coding[J].arXiv:1807.03748,2018. [39]CHAPELLE O,SCHOLKOPF B,ZIEN A.Semi-supervisedlearning[J].IEEE Transactions on Neural Networks,2009,20(3):542-542. [40]SOHN K,BERTHELOT D,CARLINI N,et al.Fixmatch:Simplifying semi-supervised learning with consistency and confidence[J].Advances in Neural Information Processing Systems,2020,33:596-608. [41]HU Z,YANG Z,HU X,et al.Simple:similar pseudo label exploitation for semi-supervised classification[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2021:15099-15108. [42]ARAZO E,ORTEGO D,ALBERT P,et al.Pseudo-labeling and confirmation bias in deep semi-supervised learning[C]//2020 International Joint Conference on Neural Networks(IJCNN).IEEE,2020:1-8. [43]TARVAINEN A,VALPOLA H.Mean teachers are better role models:Weight-averaged consistency targets improve semi-supervised deep learning results[C]//Proceedings of the 31st International Conference on Neural Information Processing Systems.2017:1195-1204. [44]LEE D H.Pseudo-label:The simple and efficient semi-supervised learning method for deep neural networks[C]//Workshop on Challenges in Representation Learning.2013:896. [45]BERTHELOT D,CARLINI N,GOODFELLOW I,et al.Mixmatch:A holistic approach to semi-supervised learning[C]//Proceedings of the 33rd International Conference on Neural Information Processing Systems.2019:5049-5059. [46]BERTHELOT D,CARLINI N,CUBUK E D,et al.Remix-match:Semi-supervised learning with distribution alignment and augmentation anchoring[J].arXiv:1911.09785,2019. [47]SELLARS P,AVILES-RIVERO A I,SCHÖNLIEB C B.Laplacenet:A hybrid energy-neural model for deep semi-supervised classification[J].arXiv:2106.04527,2021. [48]ZHANG B,WANG Y,HOU W,et al.Flexmatch:Boostingsemi-supervised learning with curriculum pseudo labeling[J].Advances in Neural Information Processing Systems,2021,34:18408-18419. [49]CHAPELLE O,SCHOLKOPF B,ZIEN A.Semi-supervisedlearning[J].IEEE Transactions on Neural Networks,2009,20(3):542-542. [50]ZHU X,GOLDBERG A B.Introduction to semi-supervisedlearning[J].Synthesis Lectures on Artificial Intelligence and Machine Learning,2009,3(1):1-130. [51]JOHNSON J,DOUZE M,JÉGOU H.Billion-scale similaritysearch with gpus[J].IEEE Transactions on Big Data,2019,7(3):535-547. [52]CUBUK E D,ZOPH B,SHLENS J,et al.Randaugment:Practical automated data augmentation with a reduced search space[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops.2020:702-703. [53]HUANG J,GONG S,ZHU X.Deep semantic clustering by partition confidence maximisation[C]//Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition.2020:8849-8858. [54]ZHOU D,BOUSQUET O,LAL T,et al.Learning with local and global consistency[C]//Proceedings of the 16th International Conference on Neural Information Processing Systems.2003:321-328. [55]ZHANG H,CISSE M,DAUPHIN Y N,et al.mixup:Beyondempirical risk minimization[J].arXiv:1710.09412,2017. [56]HE K,SUN J.Convolutional neural networks at constrainedtime cost[C]//Proceedings of the IEEE Conference on Compu-ter Vision and Pattern Recognition.2015:5353-5360. [57]KRIZHEVSKY A,HINTON G.Learning multiple layers of features from tiny images[J/OL].Handbook of Systemic Autoimmune Diseases,2009,1(4).https://scholar.google.com/scho-lar?hl=zh-CN&as_sdt=0%2C5&q=Learning+multiple+la-yers+of+features+from+tiny+image&btnG=. [58]KINGMA D P,BA J.Adam:A method for stochastic optimization[J].arXiv:1412.6980,2014. [59]WANG F,JIANG M,QIAN C,et al.Residual attention network for image classification[C]//Proceedings of the IEEE Confe-rence on Computer Vision and Pattern Recognition.2017:3156-3164. [60]DEVRIES T,TAYLOR G W.Improved regularization of convolutional neural networks with cutout[J].arXiv:1708.04552,2017. [61]MCDAID A F,GREENE D,HURLEY N.Normalized mutual information to evaluate overlapping community finding algorithms[J].arXiv:1110.2515,2011. [62]HUBERT L,ARABIE P.Comparing partitions[J].Journal of Classification,1985,2(1):193-218. [63]MACQUUEN J B.Some methods for classification and analysis of multivariate observation[C]//Proceedings of the 5th Berkley Symposium on Mathematical Statistics and Probability.1967:281-297. [64]ZELNIK-MANOR L,PERONA P.Self-tuning spectral cluste-ring[C]//Proceedings of the 17th International Conference on Neural Information Processing Systems.2004:1601-1608. [65]CAI D,HE X,WANG X,et al.Locality preserving nonnegative matrix factorization[C]//21st International Joint Conference on Artificial Intelligence(IJCAI 2009).2009:1010-1015. [66]NG A.Sparse autoencoder[J].CS294A Lecture Notes,2011,72(2011):1-19. [67]VINCENT P,LAROCHELLE H,LAJOIE I,et al.Stacked denoising autoencoders:Learning useful representations in a deep network with a local denoising criterion[J].Journal of Machine Learning Research,2010,11(12):3371-3408. [68]BENGIO Y,LAMBLIN P,POPOVICI D,et al.Greedy layer-wise training of deep networks[C]//Proceedings of the 19th International Conference on Neural Information Processing Systems.2006:153-160. [69]RADFORD A,METZ L,CHINTALA S.Unsupervised representation learning with deep convolutional generative adversarial networks[J].arXiv:1511.06434,2015. [70]KINGMA D P,WELLING M.Auto-encoding variational bayes[J].arXiv:1312.6114,2013. [71]HAEUSSER P,PLAPP J,GOLKOV V,et al.Associative deep clustering:Training a classification network with no labels[C]//German Conference on Pattern Recognition.Cham:Springer,2018:18-32. [72]CARON M,BOJANOWSKI P,JOULIN A,et al.Deep cluste-ring for unsupervised learning of visual features[C]//Procee-dings of the European Conference on Computer Vision (ECCV).2018:132-149. [73]CHANG J,GUO Y,WANG L,et al.Deep discriminative clustering analysis[J].arXiv:1905.01681,2019. |
|