摘要： 超图谱聚类方法由于能很好地描述数据点间的高阶信息,近年来受到了广泛的关注。不同于传统图结构,超图结构中的超边不是两两数据点间的连接,而是一组具有某种相同特性的数据子集。在实际应用中,常用K-近邻来构建超图中的超边,因此,并没有考虑到数据内在的关联性。提出一种新的基于稀疏重构的超图构建方法。对每一样本,用稀疏表示来找到与其最有关联的近邻样本,以此形成基于稀疏重构的超图模型,使得每个超边内的样本都具有很强的关联性。最后通过对超图拉普拉斯矩阵进行谱分解得到聚类结果。在人脸数据库、手写体数据库上的实验结果验证了算法的有效性。
| [1] Belkin M,Niyog P.Laplacian eigen mapsfor dimensionality reduc-tion and data representation [J]．Neural Comput,2002,6(15):1373-1396 [2] Roweis S,Saul L.Nonlinear dimensionality reduction by locally linear embedding [J].Science,2000,0(5500):2323-2326 [3] Tenenbaum J,Silva V,Langford J.Aglobal geometric frame-work for nonlinear dimensionality reduction [J]．Science,2000,0(5500):2319-2323 [4] Shi J B,Malik J.Motion segmentation and tracking using normalized cuts [C]∥IEEE International Conference on Computer Vision (ICCV)．1998:1154-1160 [5] Shi J B,Malik J.Normalized cuts and image segmentation [J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2000,2(8):888-905 [6] Agarwal S,Branson K,Belongie K．Higher order learning withgraphs [C]∥Int.Conf.Mach.Learn．Pittsburgh,PA,2006:17-24 [7] Agarwal S,Lim J,Zelnik M L,et al．Beyond pairwise clustering[C]∥Int．Conf．Comput．Vis．Pattern Recog．San Diego,CA,2005:838-845 [8] Yu J,Tao D C,Wang M．Adaptive Hypergraph Learning and its Application in Image Classfication [J]．IEEE Transactions on image processing,2012,21(7):3262-3272 [9] Rodr′equez J.On the laplacian spectrum and walk-regular hypergraphs [J]．Linear and Multilinear Algebra,2003,15(3):285-297 [10] Cheng B,Yang J,Yan S,et al.Learning with L1-graph for image analysis [J].IEEE Trans．Image Process,2010,9(4):858 [11] Cai D,He X F,Han J．Active subspace learning [C]∥ICCV’09:IEEE International Conference on Computer Vision．2009:911-916 [12] Huang Y,Liu Q S,Metaxas D．Video object segmentation by hypergraph cut [C]∥Int．Conf．Comput．Vis．Pattern Recog．Miami,FL,2009:1738-1745 [13] Sun L,Ji S,Ye J.Hypergraph spectral learning for multilabelclassification [C]∥Proc．Int．Conf．Know．Discov.Data Mining.Las Vegas,NV,2008:668-676 [14] Guan N,Tao D,Luo Z,et al.Manifold regularizeddiscriminative non-negative matrix factorization with fast gradient descent [J].IEEE Trans．Image Process.,2011,20(7):2030-2048 [15] Zhou D,Huang J,Schlkopf B．Learning with hypergraphs:Clustering,classification,and embedding [C]∥Neural Inf．Process．Syst.Vancouver,BC,Canada,2006:1601-1608 [16] Huang Y,Liu Q S,Lv F,et al.Unsupervised image categoryzation by hypergraph partition [J]．IEEE Trans．Pattern Anal．Mach．Intell,2011,33(6):1266-1273 [17] Wright J,Yang A,Ganesh A,et al.Robust face recognition via sparse representation [J]．IEEE Trans．on PAMI,2008,31(2):210-227 [18] Zheng X,Cai D,He X F,et al.Locality preserving clustering for image database[C]∥ACM Int．Conf．Multimedia．2004:885-891 [19] Lin Z,Liu R,Su Z．Linearized alternating direction method with adaptive penalty for low rank representation [C]∥NIPS．2011:612-620 [20] 陈丽敏,杨静,张健沛.一种基于加速迭代的大数据集谱聚类方法 [J].计算机科学,2012,39(5):172-175 [21] Huang Y,Liu Q S,Zhang S,et al.Image retrieval via probabilistichypergraph ranking [C]∥Int．Conf．Comput．Vis．Pattern Recog.San Francisco,CA,2010:3376-3383 [22] Zheng M,Bu J,Chen C,et al.Graph Regularized Sparse Coding for Image Representation [J]．IEEE Trans．Image Process,2011,20(5):1057-7149 [23] Joliffe I．Principal Component Analysis [M].New York:Springer-Verlag,1986:1580-1584 [24] Chen S,Donoho D,Saunders D．Atomicdecomposition by basis pursuit [J].Soc．Ind．Appl．Math．Rev．,2011,3(1):129-159 [25] Alpert C J,Kahng A B．Recent directions in netlist partitioning:A survey [J]．Integration:The VLSI Journal,1995,19(1/2):1-81 [26] Bolla M．Spectra euclidean representations and clustering of hypergraphs [J]．Discrete Mathematics,1993,117(1-3):19-39 [27] Zhuang L,Gao H,Lin Z,et al.Non-Negative Low Rank and Sparse Graph for Semi-Supervised Learning[C]∥Proceedings of IEEE Conference on CVPR．Providence,RI,2012:2328-2335 |
| No related articles found! |
|
||
首页