On the Quality of a Semidefinite Programming Bound for Sparse Principal Component Analysis

dc.creatorGhaoui, Laurent El
dc.date2006-01-18
dc.date2006-02-03
dc.date.accessioned2026-07-07T08:07:28Z
dc.date.available2026-07-07T08:07:28Z
dc.descriptionWe examine the problem of approximating a positive, semidefinite matrix $Σ$ by a dyad $xx^T$, with a penalty on the cardinality of the vector $x$. This problem arises in sparse principal component analysis, where a decomposition of $Σ$ involving sparse factors is sought. We express this hard, combinatorial problem as a maximum eigenvalue problem, in which we seek to maximize, over a box, the largest eigenvalue of a symmetric matrix that is linear in the variables. This representation allows to use the techniques of robust optimization, to derive a bound based on semidefinite programming. The quality of the bound is investigated using a technique inspired by Nemirovski and Ben-Tal (2002).
dc.description13 pages, 3 figures This new version corresponds to an extensive revision of the earlier version
dc.identifierhttps://arxiv.org/abs/math/0601448
dc.identifierhttp://arxiv.org/abs/math/0601448
dc.identifier.urihttp://salesiana.dossiersoluciones.com/handle/123456789/130949
dc.subjectOptimization and Control
dc.subjectStatistics Theory
dc.titleOn the Quality of a Semidefinite Programming Bound for Sparse Principal Component Analysis
dc.typetext

Files

Collections