Accurate and efficient measurement of observations is critical for scientific inquiry. There has been a growing interest by researchers in the design of adaptive experiments that lead to rapid accumulation of information about the phenomenon under study with the fewest possible measurements. Our lab has developed one such Bayesian method, dubbed adaptive design optimization (ADO). ADO currently operates under the simplifying assumption that one of the models being tested is the true, data-generating model. This assumption is most certainly violated in practice because all models are imperfect and approximate representations of the underlying system of interest. As such, ADO is not robust. We introduce a semi-parametric Bayesian method that extends ADO to make it robust to model misspecification. Specifically, two statistical tools, Bayesian penalized-splines and Bayesian variable selection, are combined with ADO. Results from preliminary simulations as well as empirical validation of the method will be discussed

Jay Myung is Professor of Psychology at the Ohio State University, USA. His research interests include cognitive science, computational modeling, Bayesian inference methods, active learning, and optimal experimental design. He did B.S. In physics from Seoul National University, M.S. In biochemistry from KAIST, PhD in Psychology from Purdue University, and postdoctoral training in Neuroscience from University of Virginia. He has been at Ohio State since 1991.