Frequentist and Bayesian Lasso Techniques for Parameter Selection in Nonlinearly Parameterized Models*

In this paper, we discuss the use of frequentist and Bayesian lasso (least absolute shrinkage and selection operator) techniques for parameter selection in nonlinearly parameterized models employed for control design. This is necessary to isolate the subset of identifiable or influential parameters, which can be uniquely calibrated from experimental data. We survey the performance of existing algorithms and present a new Bayesian lasso implementation based on the Delayed Rejection Adaptive Metropolis (DRAM) algorithm.