Abstract
The paper investigates several notions of graph Laplacians and graph kernels from the perspective of understanding the graph clustering via the graph embedding into an Euclidean space. We propose hereby a unified view of spectral graph clustering and kernel clustering methods. The various embedding techniques are evaluated from the point of view of clustering stability (with respect to k-means that is the algorithm underpinning the spectral and kernel methods). It is shown that the choice of a fixed number of dimensions may result in clustering instability due to eigenvalue ties. Furthermore, it is shown that kernel methods are less sensitive to the number of used dimensions due to downgrading the impact of less discriminative dimensions.
This is a preview of subscription content, log in via an institution to check access.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
Similar content being viewed by others
Notes
- 1.
The graphs may originate from data embedded in Euclidean space, for which a graph was constructed based on thresholding similarities/distances between the objects [12].
- 2.
\(s_{ii}=0\) is set for technical reasons. Self-similarity does not affect the basic formula, but the value 0 is useful when handling Laplacians.
- 3.
As the matrix S is symmetric, this definition is a bit too laborious, but later on we will refer to asymmetric matrices.
References
Avrachenkov, K., Chebotarev, P., Rubanov, D.: Kernels on graphs as proximity measures. In: Bonato, A., Chung Graham, F., Prałat, P. (eds.) WAW 2017. LNCS, vol. 10519, pp. 27–41. Springer, Cham (2017). https://doi.org/10.1007/978-3-319-67810-8_3
Belkin, M., Niyogi, P.: Laplacian eigenmaps for dimensionality reduction and data representation. Neural Comput. 15(6), 1373–1396 (2003)
Dhillon, I., Guan, Y., Kulis, B.: A unified view of kernel k-means, spectral clustering and graph cuts. Technical report, UTCS Technical report, TR-04-25, February 2005
Filippone, M., Camastra, F., Masulli, F., Rovetta, S.: A survey of kernel and spectral methods for clustering. Pattern Recogn. 41(1), 176–190 (2008)
Hagen, L., Kahng, A.: New spectral methods for ratio cut partitioning and clustering. IEEE Trans. Comput. Aided Des. Integr. Circuits Syst. 11(9), 1074–1085 (1992)
Hall, K.M.: An r-dimensional quadratic placement problem. Manage. Sci. 17(3), 219–229 (1970)
Hofmann, T., Schölkopf, B., Smola, A.: Kernel methods in machine learning. Ann. Stat. 36(3) (2008)
Kłopotek, M., Wierzchoń, S., Kłopotek, R.: Weighted Laplacians of grids and their application for inspection of spectral graph clustering methods (2020, submitted)
Kurras, S.: Variants of the graph Laplacian with applications in machine learning. Ph.D. thesis, University of Hamburg, Germany (2017)
von Luxburg, U.: A tutorial on spectral clustering. Stat. Comput. 17(4), 395–416 (2007)
Ng, A.Y., Jordan, M.I., Weiss, Y.: On spectral clustering: analysis and an algorithm. In: Advances in Neural Information Processing Systems, pp. 849–856. MIT Press (2002)
Schiebinger, G., Wainwright, M.J., Yu, B.: The geometry of kernelized spectral clustering. Ann. Statist. 43(2), 819–846 (2015)
Shi, J., Malik, J.: Normalized cuts and image segmentation. IEEE Trans. Pattern Anal. Mach. Intell. 22(8), 888–905 (2000)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2020 Springer Nature Switzerland AG
About this paper
Cite this paper
Wierzchoń, S.T., Kłopotek, M.A. (2020). Spectral Cluster Maps Versus Spectral Clustering. In: Saeed, K., Dvorský, J. (eds) Computer Information Systems and Industrial Management. CISIM 2020. Lecture Notes in Computer Science(), vol 12133. Springer, Cham. https://doi.org/10.1007/978-3-030-47679-3_40
Download citation
DOI: https://doi.org/10.1007/978-3-030-47679-3_40
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-030-47678-6
Online ISBN: 978-3-030-47679-3
eBook Packages: Computer ScienceComputer Science (R0)