Abstract
In this paper, we study convex optimization methods for computing the nuclear (or, trace) norm regularized least squares estimate in multivariate linear regression. The so-called factor estimation and selection method, recently proposed by Yuan et al. (J Royal Stat Soc Ser B (Statistical Methodology) 69(3):329–346, 2007) conducts parameter estimation and factor selection simultaneously and have been shown to enjoy nice properties in both large and finite samples. To compute the estimates, however, can be very challenging in practice because of the high dimensionality and the nuclear norm constraint. In this paper, we explore a variant due to Tseng of Nesterov’s smooth method and interior point methods for computing the penalized least squares estimate. The performance of these methods is then compared using a set of randomly generated instances. We show that the variant of Nesterov’s smooth method generally outperforms the interior point method implemented in SDPT3 version 4.0 (beta) (Toh et al. On the implementation and usage of sdpt3—a matlab software package for semidefinite-quadratic-linear programming, version 4.0. Manuscript, Department of Mathematics, National University of Singapore (2006)) substantially. Moreover, the former method is much more memory efficient.
Similar content being viewed by others
References
Anderson T.W.: Estimating linear restriction on regression coefficients for multivariate normal distributions. Ann. Appl. Probab. 22, 327–351 (1951)
Bach F.: Consistency of trace norm minimization. J. Mach. Learn. Res. 8, 1019–1048 (2008)
Ben-Tal, A., Nemirovski, A.: Lectures on Modern Convex Optimization: Analysis, algorithms, Engineering Applications. MPS-SIAM Series on Optimization. SIAM, Philadelphia (2001)
Bertsekas D.: Nonlinear Programming. 2nd edn. Athena Scientific, New York (1999)
Breiman L.: Heuristics of instability and stabilization in model selection. Ann. Stat. 24, 2350–2383 (1996)
Brooks R., Stone M.: Joint continuum regression for multiple predictands. J. Am. Stat. Assoc. 89, 1374–1377 (1994)
Fazel, M., Hindi, H., Boyd, S.P.: A rank minimization heuristic with application to minimum order system approximation. In: Proceedings American Control Conference, vol. 6, pp. 4734–4739 (2001)
Hiriart-Urruty J.-B., Lemaréchal C.: Convex Analysis and Minimization algorithms I, volume 305 of Comprehensive Study in Mathematics. Springer, New York (1993)
Hotelling H.: The most predictable criterion. J. Educational Psychol. 26, 139–142 (1935)
Hotelling H.: Relations between two sets of variables. Biometrika 28, 321–377 (1936)
Izenman A.: Reduced-rank regression for the multivariate linear model. J. Multivar. Anal. 5, 248–264 (1975)
Lu Z.: Smooth optimization approach for covariance selection. SIAM J. Optim. 19, 1807–1827 (2009)
Lu, Z., Monteiro, R.D.C., Yuan, M.: Convex optimization methods for dimension reduction and coefficient estimation in multivariate linear regression. Technical report, Department of Mathematics, Simon Fraser University, Burnaby, BC, V5A 1S6, Canada, January 2008
Massy W.: Principle components regression with exploratory statistical research. J. Am. Stat. Assoc. 60, 234–246 (1965)
Nesterov, Y. E.: A method for unconstrained convex minimization problem with the rate of convergence O(1/k 2). Doklady AN SSSR, 269:543–547, 1983. Translated as Soviet Math. Docl
Nesterov Y.E.: Smooth minimization of nonsmooth functions. Math. Program. 103, 127–152 (2005)
Recht, B., Fazel, M., Parrilo, P.A. (2007) Guaranteed minimum-rank solutions of linear matrix equations via nuclear norm minimization. Technical report arXiv:0706.4138v1, arXiv
Reinsel G., Velu R.: Multivariate Reduced-rank Regression: Theory and Application. Springer, New York (1998)
Tibshirani R.: Regression shrinkage and selection via the lasso. J. Royal. Statist. Soc. B 58, 267–288 (1996)
Toh, K.C., Tütüncü, R.H., Todd, M.J.: On the implementation and usage of sdpt3—a matlab software package for semidefinite-quadratic-linear programming, version 4.0. Manuscript, Department of Mathematics, National University of Singapore, July 2006
Tseng, P.: On accelerated proximal gradient methods for convex–concave optimization. Manuscript, Department of Mathematics, University of Washington, May 2008
van den Berg, E., Friedlander, M.: A root-finding approach for sparse recovery. Talk given at ISMP, Chicago, 23–28 August 2009
van den Berg, E., Friedlander, M.: In pursuit of a root. Working paper, Department of Computer Science, University of British Columbia, November 2009
Wold, H.: Soft modeling by latent variables: the nonlinear iterative partial least squares approach. In: In Perspectives in Probability and Statistics: Papers in Honor of M. S. Bartlett. Academic Press, New York (1975)
Yuan M., Ekici A., Lu Z., Monteiro R.D.C.: Dimension reduction and coefficient estimation in multivariate linear regression. J. R. Stat. Soc. Series B Stat. Methodol. 69(3), 329–346 (2007)
Author information
Authors and Affiliations
Corresponding author
Additional information
Zhaosong Lu was supported in part by SFU President’s Research Grant and NSERC Discovery Grant. Renato D. C. Monteiro was supported in part by NSF Grants CCF-0430644, CCF-0808863 and CMMI-0900094 and ONR Grants N00014-05-1-0183 and N00014-08-1-0033. Ming Yuan was supported in part by NSF Grants DMS-0624841 and DMS-0706724.
Rights and permissions
About this article
Cite this article
Lu, Z., Monteiro, R.D.C. & Yuan, M. Convex optimization methods for dimension reduction and coefficient estimation in multivariate linear regression. Math. Program. 131, 163–194 (2012). https://doi.org/10.1007/s10107-010-0350-1
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10107-010-0350-1
Keywords
- Cone programming
- Smooth saddle point problem
- First-order method
- Multivariate linear regression
- Nuclear or trace norm
- Dimension reduction