Speaker

   Prof. Yu Ding (Texas A&M University) 

　　Title

   Gaussian Process Models: Fast Computation and Engineering Applications                        

　　Time

    2013.7.4 10:00am                                      

　　Venue

　　S703

　　Abstract

    Gaussian process (GP) regression is a flexible and powerful tool in machine learning. One critical shortcoming is that GP models do not scale well to handle high volume data. We recently develop a new approach for fast computation of Gaussian process regression with a focus on large spatial data sets. Our approach decomposes the domain of a regression function into small subdomains and infers a local piece of the regression function for each subdomain. We then stitch all the local pieces together to make a coherent, globally connected response function. This new approach entertains a couple of advantages: it is faster than the alternatives, can be parallelized easily, making it even faster, and it can be adaptive to non-stationary features in the data, because of its use of different parameters for individual local regions. In this talk, I’ll demonstrate the advantages of the new method using some NASA satellite data. At the end of the talk, I shall discuss how the GP models have been used in a number of our recent engineering applications.

　　Affiliation

     Dr. Yu Ding is currently a Professor of Industrial & Systems Engineering and a Professor of Electrical & Computer Engineering, as well as a Faculty Affiliate with the Institute of Applied Mathematics and Computational Sciences (IAMCS), all at Texas A&M University. Dr. Ding received a B.S. degree from the University of Science & Technology of China in 1993, an M.S. degree from Tsinghua University in 1996, an M.S. degree from Penn State University in 1998, and a Ph.D. degree from the University of Michigan in 2001. His research interests are in the general areas of system informatics, and quality and reliability engineering. Dr. Ding currently serves as a department editor for IIE Transactions. He is senior member of IEEE, and a member of IIE, INFORMS and ASME. More information is available on his Lab’s website, http://ise.tamu.edu/metrology.