6533b82afe1ef96bd128b824

RESEARCH PRODUCT

Non-linear System Identification with Composite Relevance Vector Machines

Manel Martínez-ramónJordi Munoz-mariJosé Luis Rojo-álvarezGustau Camps-valls

subject

Relevance Vector MachinesTelecomunicacionesNonlinear system identificationbusiness.industryRVMApplied MathematicsNonlinear System IdentificationRegression analysisPattern recognitionComposite kernelsFunction (mathematics)Support vector machineNonlinear systemStatistics::Machine LearningSignal ProcessingBenchmark (computing)3325 Tecnología de las TelecomunicacionesRelevance (information retrieval)Artificial intelligenceElectrical and Electronic EngineeringbusinessMathematicsFree parameter

description

Nonlinear system identification based on relevance vector machines (RVMs) has been traditionally addressed by stacking the input and/or output regressors and then performing standard RVM regression. This letter introduces a full family of composite kernels in order to integrate the input and output information in the mapping function efficiently and hence generalize the standard approach. An improved trade-off between accuracy and sparsity is obtained in several benchmark problems. Also, the RVM yields confidence intervals for the predictions, and it is less sensitive to free parameter selection. Teoría de la Señal y Comunicaciones

yearjournalcountryeditionlanguage
2007-04-01
10.1109/lsp.2006.885290https://hdl.handle.net/10016/11671