基于超图正则化非负Tucker分解的图像聚类算法

doi:10.19678/j.issn.1000-3428.0060601

摘要/Abstract

摘要： 针对非负张量分解应用于图像聚类时忽略了高维数据内部几何结构的问题，在经典的张量非负Tucker分解的基础上，添加超图正则项以尽可能多地保留原始数据的内在几何结构信息，提出一种基于超图正则化非负Tucker分解模型HGNTD。通过构造超图刻画数据内部样本间的高阶关系，提高几何结构描述的准确性，针对超图正则化非负张量分解模型，基于交替非负最小二乘法，设计快速有效的超图正则化非负Tucker分解算法求解所给模型，证明算法在非负的条件下是收敛的，最终将算法应用于图像聚类。在Yale和COIL两个常用公开数据集上的实验结果表明，相对于k-means、非负矩阵分解、图正则化非负矩阵分解、非负Tucker分解和图正则化非负Tucker分解等算法，超图正则化非负Tucker分解算法聚类准确度提升了8.6%~11.4%，归一化互信息提升了2.0%~7.5%，具有更好的聚类效果。

关键词: 非负张量分解, Tucker分解, 超图学习, 交替非负最小二乘法, 聚类分析

Abstract: The internal geometry structure of high-dimensional data is ignored when nonnegative tensor decomposition is applied to image clustering.To solve this problem, we propose a Hypergraph regularized Nonnegative Tucker Decomposition(HGNTD) model by adding a hypergraph regularization term to preserve the intrinsic geometric structure information of original data as much as possible.This method is based on the classical nonnegative Tucker decomposition of tensors.Specifically, relying on the hypergraph regularization non-negative tensor decomposition model and based on the Alternating Nonnegative Least Squares(ANLS), a fast and effective hypergraph regularized nonnegative tucker decomposition algorithm is designed to solve the given model.The new algorithm is proven to converge under nonnegative conditions, thereby, it is applied to image clustering.The experimental results on Yale and COIL-100 data sets show that, compared with k-means, Nonnegative Matrix Factorization(NMF), Graph regularized Nonnegative Matrix Factorization(GNMF), Nonnegative Tucker Decomposition(NTD), and Graph regularized Nonnegative Tucker Decomposition(GNTD) algorithms, the clustering accuracy of the new algorithm is improved by approximately 8.6%~11.4%, and the Normalized Mutual Information(NMI) is improved by approximately 2.0%~7.5%.Therefore, it can be said to have a better clustering effect.

Key words: nonnegative tensor decomposition, Tucker decomposition, hypergraph learning, alternating nonnegative least squares, clustering analysis

中图分类号:

TP391

陈璐瑶, 刘奇龙, 许云霞, 陈震. 基于超图正则化非负Tucker分解的图像聚类算法[J]. 计算机工程, 2022, 48(4): 197-205.

CHEN Luyao, LIU Qilong, XU Yunxia, CHEN Zhen. Image Clustering Algorithm Based on Hypergraph Regularized Nonnegative Tucker Decomposition[J]. Computer Engineering, 2022, 48(4): 197-205.

http://www.ecice06.com/CN/Y2022/V48/I4/197

图/表 11

20230131201731

20230131201734

20230131201737

20230131201741

20230131201744

20230131201747

20230131201750

20230131201753

20230131201756

20230131201759

20230131201802

参考文献

[1] VASWANI N, BOUWMANS T, JAVED S, et al.Robust subspace learning:robust PCA, robust subspace tracking, and robust subspace recovery[EB/OL].[2020-12-10].https://arxiv.org/pdf/1711.09492.pdf.
[2] LIU J, OSHER S.Block matching local SVD operator based sparsity and TV regularization for image denoising[J].Journal of Scientific Computing, 2019, 78(1):607-624.
[3] SAFONT G, SALAZAR A, VERGARA L, et al.Probabilistic distance for mixtures of independent component analyzers[J].IEEE Transactions on Neural Networks and Learning Systems, 2018, 29(4):1161-1173.
[4] FU X, HUANG K, SIDIROPOULOS N D, et al.Nonnegative matrix factorization for signal and data analytics:identifiability, algorithms, and applications[EB/OL].[2020-12-10].https://arxiv.org/pdf/1803.01257.pdf.
[5] LEE D D, SEUNG H S.Learning the parts of objects by non-negative matrix factorization[J].Nature, 1999, 401(6755):788-791.
[6] YAN S, XU D, YANG Q, et al.Multilinear discriminant analysis for face recognition[J].IEEE Transactions on Image Processing, 2007, 16(1):212-220.
[7] CICHOCKI A, MANDIC D, DE LATHAUWER L, et al.Tensor decompositions for signal processing applications from two-way to multiway component analysis[J].IEEE Signal Processing Magazine, 2015, 32(2):145-163.
[8] HARSHMAN R A.Foundations of the parafac procedure:models and conditions for an "explanatory" multimodal factor analysis[EB/OL].[2020-12-10].https://psychology.uwo.ca/faculty/harshman/wpppfac0.pdf.
[9] TUCKER L R.Some mathematical notes on three-mode factor analysis[J].Psychometrika, 1966, 31(3):279-311.
[10] KOLDA T G, TAMARA G, BADER, et al.Tensor decompositions and applications[J].SIAM Review, 2009, 51(3):455-500.
[11] ZHOU G, CICHOCKI A, ZHAO Q, et al.Efficient nonnegative tucker decompositions:algorithms and uniqueness[J].IEEE Transactions on Image Processing, 2015, 24(12):4990-5003.
[12] LI S Z, HOU X W, ZHANG H J, et al.Learning spatially localized, parts-based representation[C]//Proceedings of 2001 IEEE Computer Society Conference on Computer Vision & Pattern Recognition.Kauai, USA:IEEE Press, 2001:1582-1598.
[13] CAI D, HE X, WU X, et al.Non-negative matrix factorization on manifold[C]//Proceedings of the 8th IEEE International Conference on Data Mining.Washington D.C., USA:IEEE Press, 2008:63-72.
[14] CAI D, HE X, HAN J, et al.Graph regularized nonnegative matrix factorization for data representation[J].IEEE Transactions on Pattern Analysis and Machine Intelligence, 2011, 33(8):1548-1560.
[15] LENG C, ZHANG H, CAI G, et al.Graph regularized LP smooth non-negative matrix factorization for data representation[J].IEEE/CAA Journal of Automatica Sinica, 2019, 6(2):584-595.
[16] WANG Y, GUI L, ZHANG Y.Neighborhood preserving non-negative tensor factorization for image representation[C]//Proceedings of IEEE International Conference on Acoustics.Washington D.C., USA:IEEE Press, 2012:3889-3392.
[17] WANG C, HE X, BU J, et al.Image representation using Laplacian regularized nonnegative tensor factorization[J].Pattern Recognition, 2011, 44(10/11):2516-2526.
[18] KIM Y D, CHOI S.Nonnegative tucker decomposition[C]//Proceedings of IEEE Conference on Computer Vision & Pattern Recognition.Washington D.C., USA:IEEE Press, 2008:1-8.
[19] JIANG B, DING C, TANG J, et al.Image representation and learning with graph-Laplacian tucker tensor decomposition[J].IEEE Transactions on Cybernetics, 2019, 49(4):1417-1426.
[20] LI X, NG M K, CONG G, et al.MR-NTD:manifold regularization nonnegative tucker decomposition for tensor data dimension reduction and representation[J].IEEE Transactions on Neural Networks and Learning Systems, 2017, 28(8):1787-1800.
[21] QIU Y, ZHOU G, ZHANG Y, et al.Graph regularized nonnegative tucker decomposition for tensor data representation[C]//Proceedings of IEEE International Conference on Acoustics, Speech and Signal Processing.Washington D.C., USA:IEEE Press, 2019:8613-8617.
[22] ZHOU D, HUANG J, SCHLKOPF B.Learning with hypergraphs:clustering, classification, and embedding[C]//Proceedings of the 20th Annual Conference on Neural Information Processing Systems.Vancouver, Canada:MIT Press, 2006:134-146.
[23] AGARWAL S, BRANSON K, BELONGIE S.Higher order learning with graphs[C]//Proceedings of the 23rd International Conference on Machine Learning.Pittsburgh, USA:[s.n.], 2006:121-128.
[24] LIU M, ZHANG J, GUO X, et al.Hypergraph regularized sparse feature learning[J].Neurocomputing, 2017, 237(10):185-192.
[25] ZENG K, YU J, LI C, et al.Image clustering by hyper-graph regularized non-negative matrix factorization[J].Neurocomputing, 2014, 138(22):209-217.
[26] WANG W, QIAN Y, TANG Y Y.Hypergraph-regularized sparse NMF for hyperspectral unmixing[J].IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, 2016, 9(2):681-694.
[27] 罗永恩, 胡继承, 徐茜.基于超图的多模态关联特征处理方法[J].计算机工程, 2017, 43(1):226-230. LUO Y E, HU J C, XU Q.Multimodal correlation feature processing method based on hypergraph[J].Computer Engineering, 2017, 43(1):226-230.(in Chinese)
[28] BELKIN M, NIYOGI P.Laplacian eigenmaps and spectral techniques for embedding and clustering[J].Advances in Neural Information Processing Systems, 2001, 14(6):585-591.
[29] LEE D D, SEUNG H S.Algorithms for non-negative matrix factorization[C]//Proceedings of International Conference on Neural Information Processing Systems.[S.1.]:MIT Press, 2000:135-153.
[30] MACQUEEN J B.Some methods for classification and analysis of multivariate observations[C]//Proceedings of the 5th Berkeley Symposium on Mathematical Statistics and Probability.Los Angeles, USA:University of California Press, 1967:238-247.
[31] XU W, LIU X, GONG Y.Document clustering based on non-negative matrix factorization[C]//Proceedings of the 26th Annual International ACM SIGIR Conference on Research and Development in Information Retrieval.New York, USA:ACM Press, 2003:267-273.

选择文件类型/文献管理软件名称

选择包含的内容