Nonparametric Machine Learning and Bayesian Nonparametrics

Distributed Autonomous System Laboratory

Project Description: 

Nonparametric machine learning is concerned with building predictive models of stochastic phenomena without assuming a prior structure.  These models provide significant flexibility in modeling time-varying and uncertain events without having to make rigid assumptions about the parametric form of the model. We pursue research in two major classes of nonparametric models: Reproducing Kernel Hibert Space (RKHS ) based, and Bayesian Nonparametric (BNP) models.

Useful resources:

A collection of presentations, papers, and MATLAB software on Gaussian Processes regression, and GP based Adaptive Control 

Planning under Uncertainty using Nonparametric Bayesian Models

GP-NBC code can be obtained for experimentation, please contact Girish Chowdhary for details.