English

Penalized versus constrained generalized eigenvalue problems

Computation 2021-04-01 v3 Machine Learning

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.

Cite

@article{arxiv.1410.6131,
  title  = {Penalized versus constrained generalized eigenvalue problems},
  author = {Irina Gaynanova and James Booth and Martin T. Wells},
  journal= {arXiv preprint arXiv:1410.6131},
  year   = {2021}
}

Comments

18 pages, 8 figures

R2 v1 2026-06-22T06:33:07.822Z