Demystify Lindley's Paradox and The Delaunay Triangulation Learner

Speaker:Guosheng Yin (HKU)
Time: Mon 09:30-10:30, 2019-12-09
Venue:Conference Room 3, 2nd floor of Jin Chun Yuan West Building


This talk contains two parts. In the first part, we revisit Lindley's paradox (Lindley, 1957) and demystify the conflicting results between Bayesian and frequentist hypothesis testing procedures based on a new notion of two-sided posterior probability. We propose to cast a two-sided hypothesis as a combination of two one-sided hypotheses along the opposite directions. The second part presents a new machine learning method, called the Delaunay triangulation learner (DTL), which is motivated from discrete differential geometry. Delaunay triangulation is often applied as a surface reconstruction algorithm that can accurately describe the properties of a surface using a linear interpolation function. (The first part is joint work with Haolun Shi at Simon Fraser University)


Guosheng Yin is now Patrick S C Poon Professor and Head of Department of Statistics and Actuarial Science of the University of Hong Kong. He received his PhD from University of North Carolina at Chapel Hill in 2003.
His research is mainly focusing on clinical trial design, survival analysis and Bayesian methods. He is the Elected Member of International Statistical Institute and Fellow of American Statistical Association (2013), also the World's top 1% of scientists by Thomson Reuters (2015). He is the associated editor many well-known international magazines, including Journal of American Statistical Association, Contemporary Clinical Trials,and etc. At present, he has published over 150 papers in the top international journals like JASA, Annals of Statistics, Biometrics and JRSSC.