Objetivo Interior point algorithms and a dramatic growth in computing power have revolutionized optimization inthe last two decades. Highly nonlinear problems which were previously thought intractable are nowroutinely solved at reasonable scales. Semidefinite programs (i.e. linear programs on the cone of positivesemidefinite matrices) are a perfect example of this trend: reasonably large, highly nonlinear but convexeigenvalue optimization problems are now solved efficiently by reliable numerical packages. This in turnmeans that a wide array of new applications for semidefinite programming have been discovered,mimicking the early development of linear programming. To cite only a few examples, semidefiniteprograms have been used to solve collaborative filtering problems (e.g. make personalized movierecommendations), approximate the solution of combinatorial programs, optimize the mixing rate ofMarkov chains over networks, infer dependence patterns from multivariate time series or produce optimalkernels in classification problems.These new applications also come with radically different algorithmic requirements. While interior pointmethods solve relatively small problems with a high precision, most recent applications of semidefiniteprogramming in statistical learning for example form very large-scale problems with comparatively lowprecision targets, programs for which current algorithms cannot form even a single iteration. Thisproposal seeks to break this limit on problem size by deriving reliable first-order algorithms for solvinglarge-scale semidefinite programs with a significantly lower cost per iteration, using for examplesubsampling techniques to considerably reduce the cost of forming gradients.Beyond these algorithmic challenges, the proposed research will focus heavily on applications of convexprogramming to statistical learning and signal processing theory where optimization and duality resultsquantify the statistical performance of coding or variable selection algorithms for example. Finally,another central goal of this work will be to produce efficient, customized algorithms for some keyproblems arising in machine learning and statistics. Ámbito científico engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringsignal processinghumanitiesartsmodern and contemporary artcinematographynatural sciencescomputer and information sciencesartificial intelligencemachine learning Programa(s) FP7-IDEAS-ERC - Specific programme: "Ideas" implementing the Seventh Framework Programme of the European Community for research, technological development and demonstration activities (2007 to 2013) Tema(s) ERC-SG-PE1 - ERC Starting Grant - Mathematical foundations Convocatoria de propuestas ERC-2010-StG_20091028 Consulte otros proyectos de esta convocatoria Régimen de financiación ERC-SG - ERC Starting Grant Institución de acogida CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE CNRS Aportación de la UE € 1 148 460,00 Dirección RUE MICHEL ANGE 3 75794 Paris Francia Ver en el mapa Región Ile-de-France Ile-de-France Paris Tipo de actividad Research Organisations Contacto administrativo Julie Zittel (Ms.) Investigador principal Alexandre Werner Geoffroy Gobert D'aspremont Lynden (Prof.) Enlaces Contactar con la organización Opens in new window Sitio web Opens in new window Coste total Sin datos Beneficiarios (1) Ordenar alfabéticamente Ordenar por aportación de la UE Ampliar todo Contraer todo CENTRE NATIONAL DE LA RECHERCHE SCIENTIFIQUE CNRS Francia Aportación de la UE € 1 148 460,00 Dirección RUE MICHEL ANGE 3 75794 Paris Ver en el mapa Región Ile-de-France Ile-de-France Paris Tipo de actividad Research Organisations Contacto administrativo Julie Zittel (Ms.) Investigador principal Alexandre Werner Geoffroy Gobert D'aspremont Lynden (Prof.) Enlaces Contactar con la organización Opens in new window Sitio web Opens in new window Coste total Sin datos