An Inverse Power Method for Nonlinear Eigenproblems with Applications in 1-Spectral Clustering and Sparse PCA

Many problems in machine learning and statistics can be formulated as (generalized) eigenproblems. In terms of the associated optimization problem, computing linear eigenvectors amounts to finding critical points of a quadratic function subject to quadratic constraints. In this paper we show that a certain class of constrained optimization problems with nonquadratic objective and constraints can be understood as nonlinear eigenproblems. We derive a generalization of the inverse power method which is guaranteed to converge to a nonlinear eigenvector. We apply the inverse power method to 1-spectral clustering and sparse PCA which can naturally be formulated as nonlinear eigenproblems. In both applications we achieve state-of-the-art results in terms of solution quality and runtime. Moving beyond the standard eigenproblem should be useful also in many other applications and our inverse power method can be easily adapted to new problems.

[1]  Ulrike von Luxburg,et al.  A tutorial on spectral clustering , 2007, Stat. Comput..

[2]  F. Clarke Optimization And Nonsmooth Analysis , 1983 .

[3]  Arthur D. Szlam,et al.  Total variation and cheeger cuts , 2010, ICML 2010.

[4]  K. Schittkowski,et al.  NONLINEAR PROGRAMMING , 2022 .

[5]  Alexandre d'Aspremont,et al.  Optimal Solutions for Sparse Principal Component Analysis , 2007, J. Mach. Learn. Res..

[6]  Jorge Cadima Departamento de Matematica Loading and correlations in the interpretation of principle compenents , 1995 .

[7]  Robert H. Halstead,et al.  Matrix Computations , 2011, Encyclopedia of Parallel Computing.

[8]  Tom Goldstein,et al.  The Split Bregman Method for L1-Regularized Problems , 2009, SIAM J. Imaging Sci..

[9]  Marc Teboulle,et al.  Fast Gradient-Based Algorithms for Constrained Total Variation Image Denoising and Deblurring Problems , 2009, IEEE Transactions on Image Processing.

[10]  Joachim M. Buhmann,et al.  Expectation-maximization for sparse and non-negative PCA , 2008, ICML '08.

[11]  Gert R. G. Lanckriet,et al.  Sparse eigen methods by D.C. programming , 2007, ICML '07.

[12]  Kung-Ching Chang,et al.  Variational methods for non-differentiable functionals and their applications to partial differential equations , 1981 .

[13]  Zhou Wei,et al.  Generalized Euler identity for subdifferentials of homogeneous functions and applications , 2008 .

[14]  Yurii Nesterov,et al.  Generalized Power Method for Sparse Principal Component Analysis , 2008, J. Mach. Learn. Res..

[15]  U. Feige,et al.  Spectral Graph Theory , 2015 .

[16]  Shai Avidan,et al.  Spectral Bounds for Sparse PCA: Exact and Greedy Algorithms , 2005, NIPS.

[17]  I. Jolliffe,et al.  A Modified Principal Component Technique Based on the LASSO , 2003 .

[18]  丸山 徹 Convex Analysisの二,三の進展について , 1977 .

[19]  Rodney Josué Biezuner,et al.  Computing the first eigenvalue of the p-Laplacian via the inverse power method. , 2009 .

[20]  Saïd Amghibech,et al.  Eigenvalues of the Discrete p-Laplacian for Graphs , 2003, Ars Comb..

[21]  R. Tibshirani,et al.  Sparse Principal Component Analysis , 2006 .

[22]  Matthias Hein,et al.  Spectral clustering based on the graph p-Laplacian , 2009, ICML '09.

[23]  Heng Tao Shen,et al.  Principal Component Analysis , 2009, Encyclopedia of Biometrics.