Nonparametric inference refers to statistical techniques that use data to infer unknown quantities of interest while making as few assumptions as possible. Typically, this involves working with large and flexible infinite-dimensional statistical models. The flexibility and adaptivity provided by nonparametric techniques is especially valuable in modern statistical problems of the current era of massive and complex datasets.
Berkeley statistics faculty work on many aspects of nonparametric inference. Current research interests include nonparametric hypothesis testing based on ranks and permutations, nonparametric regression, classification and density estimation under smoothness and shape constraints, high-dimensional nonparametric inference, theoretical analysis of nonparametric procedures and applications to biological research. Berkeley statistics is also a major center for the study of Bayesian methods for nonparametric inference and their applications to various areas in machine learning.
uncertainty quantification and inference, inverse problems, nonparametrics, risk assessment, earthquake prediction, election auditing, geomagnetism, cosmology, litigation, food/nutrition