LIP6 2012/001

• Reports «Méthode de Plans Sécants Régularisée pour l'Optimisation Non convexe : Annexe à l'article dans JMLR»
• T. M. T. Do, Th. Artières
• 15 pages - 11/14/2012 - document en - http://www.lip6.fr/lip6/reports/2012/lip6-2012-001.pdf 314 Ko
• Contact Thierry.Artieres (at) nulllip6.fr
• TeamMALIRE
This report is an additional material to our article in the Journal of Machine Learning Research (JMLR). Both documents deal with an algorithm that we designed, named NRBM (Non convex Regularized Bundle Methods), to deal efficiently with regularized non convex risks as often encountered in the machine learning field. The JMLR article provides more details on NRBM and report empirical evaluation on many real machine learning problems while this report provides additional theoretical results related to the convergence analysis of algorithm NRBM.