56
9
v1v2v3 (latest)

Penalized versus constrained generalized eigenvalue problems

Abstract

We investigate the difference between using an 1\ell_1 penalty versus an 1\ell_1 constraint in generalized eigenvalue problems, such as principal component analysis and discriminant analysis. Our main finding is that an 1\ell_1 penalty may fail to provide very sparse solutions; a severe disadvantage for variable selection that can be remedied by using an 1\ell_1 constraint. Our claims are supported both by empirical evidence and theoretical analysis. Finally, we illustrate the advantages of an 1\ell_1 constraint in the context of discriminant analysis and principal component analysis.

View on arXiv
Comments on this paper