Abstract
This chapter presents some practical algorithms of spectral clustering for large-scale data. Spectral clustering is a kernel-based method of grouping data on separate nonlinear manifolds. Reducing its computational expense without critical loss of accuracy contributes to its practical use especially in vision-based applications. The present algorithms exploit random projection and subsampling techniques for reducing dimensionality and the cost for evaluating pairwise similarities of data. The computation time is quasilinear with respect to the data cardinality, and it can be independent of data dimensionality in some appearance-based applications. The efficiency of the algorithms is demonstrated in appearance-based image/video segmentation.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Notes
- 1.
C can be a forward circulant matrix. We prefer the back-circulant matrix just because it is symmetric.
References
Achlioptas, D.: Database-friendly random projections: Johnson–Lindenstrauss with binary coins. J. Comput. Syst. Sci. 66, 671–687 (2003)
Ali, S., Shah, M.: A Lagrangian particle dynamics approach for crowd flow segmentation and stability analysis. In: CVPR (2007)
Berry, M.W.: Large scale sparse singular value computations. Int. J. Supercomput. Appl. 6, 13–49 (1992)
Brigham, E., Maninila, H.: Random projection in dimensionality reduction: applications to image and text data. In: ACM SIGKDD ICKDDM, pp. 245–250. ACM, New York (2001)
Carnegie Mellon University Informedia Project: Mountain Skywater, segment 11 of 12. http://www.open-video.org/ (1996)
Cour, T., Benezit, F., Shi, J.: Spectral segmentation with multiscale graph decomposition. In: CVPR Proc., vol. 2, pp. 1124–1131. IEEE Comput. Soc., Washington (2005)
Dasgupta, S., Gupta, A.: An elementary proof of the Johnson–Lindenstrauss lemma. Technical Report, UC Berkeley (1999)
Dhillon, I.S.: Co-clustering documents and words using bipartite spectral graph partitioning. In: ACM SIGKDD, pp. 269–274. ACM, New York (2001)
Ding, C.H.Q., He, X., Zha, H., Gu, M., Simon, H.D.: A min-max cut algorithm for graph partitioning and data clustering. In: Proc. of ICDM 2001, pp. 107–114 (2001)
Drineas, P., Mahoney, M.W.: On the Nyström method for approximating a Gram matrix for improved kernel-based learning. J. Mach. Learn. Res. 6, 2153–2175 (2005)
Eibl, G., Brändle, N.: Evaluation of clustering methods for finding dominant optical flow fields in crowded scenes. In: ICPR08 (2008)
Fiedler, M.: A property of eigenvectors of nonnegative symmetric matrices and its application to graph theory. Czechoslov. Math. J. 25, 619–633 (1975)
Fowlkes, C., Belongie, S., Chung, F., Malik, J.: Spectral grouping using the Nyström method. IEEE Trans. Pattern Anal. Mach. Intell. 26(2), 214–225 (2004)
Fradkin, D., Madigan, D.: Experiments with random projections for machine learning. In: ACM SIGKDD ICKDDM, pp. 517–522. ACM, New York (2003)
Freitas, N.D., Wang, Y., Mahdaviani, M., Lang, D.: Fast Krylov methods for N-body learning. In: Advances in Neural Information Processing Systems, vol. 18, pp. 251–258. MIT Press, Cambridge (2006)
Golub, G.H., Loan, C.F.V.: Matrix Computations, 3rd edn. Johns Hopkins University Press, Baltimore (1996)
Gu, M., Eisenstat, S.C.: A stable and fast algorithm for updating the singular value decomposition. Technical Report YALEU/DCS/RR-966, Yale University (1994)
Hagen, L., Kahng, A.: New spectral methods for ratio cut partitioning and clustering. IEEE Trans. Comput. Aided Des. 11(9), 1074–1085 (1992)
IEICE Information and System Society: PRMU algorithm contest 2006: Shot boundary detection from image sequence, sample data. http://www-sens.sys.es.osaka-u.ac.jp/alcon/data/level3.avi (2006) (in Japanese)
Johnson, W., Lindenstrauss, J.: Extensions of Lipschitz maps into a Hilbert space. Contemp. Math. 26, 189–206 (1984)
Mahadevan, S.: Fast spectral learning using Lanczos eigenspace projections. In: AAAI, pp. 1472–1475 (2008)
Ng, A.Y., Jordan, M.I., Weiss, Y.: On spectral clustering: analysis and an algorithm. In: Advances in Neural Information Processing Systems, vol. 14, pp. 849–856. MIT Press, Cambridge (2002)
Sakai, T.: Monte Carlo subspace method: an incremental approach to high-dimensional data classification. In: International Conference on Pattern Recognition (2008)
Scott, G.L., Longuet-Higgins, H.C.: Feature grouping by relocalisation of eigenvectors of the proximity matrix. In: British Machine Vision Conference, pp. 103–108 (1990)
Shi, J., Malik, J.: Normalized cuts and image segmentation. IEEE Trans. Pattern Anal. Mach. Intell. 22(8), 888–905 (2000)
Song, Y., Chen, W.Y., Bai, H., Lin, C.J., Chang, E.Y.: Parallel spectral clustering. In: ECML PKDD. Lecture Notes in Computer Science, vol. 5212, pp. 374–389. Springer, Berlin (2008)
Strehl, A., Ghosh, J.: Cluster ensembles—a knowledge reuse framework for combining multiple partitions. J. Mach. Learn. Res. 3, 583–617 (2002)
Vempala, S.S.: The Random Projection Method. Series in Discrete Mathematics and Theoretical Computer Science, vol. 65. American Mathematical Society, Providence (2004)
Von Luxburg, U.: A tutorial on spectral clustering. Stat. Comput. 17(4), 395–416 (2007)
Watanabe, T., Takimoto, E., Amano, K., Maruoka, A.: Random projection and its application to learning. In: Proc. 2005 Workshop on Randomness and Computation, pp. 3–4 (2005)
Weiss, Y.: Segmentation using eigenvectors: a unifying view. In: International Conference on Computer Vision, pp. 975–982 (1999)
Williams, C.K.I., Seeger, M.: Using the Nyström method to speed up kernel machines. In: Advances in Neural Information Processing Systems, vol. 13, pp. 682–688. MIT Press, Cambridge (2001)
Yu, S.X., Shi, J.: Multiclass spectral clustering. In: International Conference on Computer Vision, pp. 313–319 (2003)
Zach, C., Pock, T., Bischof, H.: A duality based approach for realtime TV-L1 optical flow. In: Pattern Recognition, Proc. DAGM, Heidelberg, Germany, pp. 214–223 (2007)
Zelnik-Manor, L., Perona, P.: Self-tuning spectral clustering. In: Advances in Neural Information Processing Systems, vol. 17, pp. 1601–1608. MIT Press, Cambridge (2004)
Zhang, K., Kwok, J.T.: Density-weighted Nyström method for computing large kernel eigensystems. Neural Comput. 21(1), 121–146 (2009). doi:10.1162/neco.2009.11-07-651
Acknowledgement
The first author was partially supported by the Grant-in-Aid for Young Scientists, from the Ministry of Education, Culture, Sports, Science and Technology of Japan under MEXT KAKEN 22700163.
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Appendix: Clustering Scores
Appendix: Clustering Scores
The conditional entropy (CE) and the normalized mutual information (NMI) [27] are defined as follows.
Here, |C i | and |A j | are the numbers of samples in the estimated cluster C i and the optimal cluster A j , respectively. X ij =C i ∩A j is the set of common samples. The smaller the CE is, or the larger the NMI is, the better the clustering result is. The NMI takes a value between 0 and 1.
Rights and permissions
Copyright information
© 2011 Springer-Verlag London Limited
About this chapter
Cite this chapter
Sakai, T., Imiya, A. (2011). Practical Algorithms of Spectral Clustering: Toward Large-Scale Vision-Based Motion Analysis. In: Wang, L., Zhao, G., Cheng, L., Pietikäinen, M. (eds) Machine Learning for Vision-Based Motion Analysis. Advances in Pattern Recognition. Springer, London. https://doi.org/10.1007/978-0-85729-057-1_1
Download citation
DOI: https://doi.org/10.1007/978-0-85729-057-1_1
Publisher Name: Springer, London
Print ISBN: 978-0-85729-056-4
Online ISBN: 978-0-85729-057-1
eBook Packages: Computer ScienceComputer Science (R0)