Speaker: Dr. Lei Cheng
Time: 10:00-12:00, Nov. 24
Location: SIST 1C 101
Host: Prof. Ziping Zhao
Our world is full of data, and these data often appear in high-dimensional structures, in which each object is described by multiple attributes. Examples include data in social sciences, medicines, pharmacology and environmental monitoring, just to name a few. To make sense of the multi-dimensional data, advanced computational tools are needed to figure out the hidden patterns underlying the data. This is where tensor models come into play. Due to the remarkable representation capability, tensor models have led to state-of-the-art performance in many domains including social network mining, image processing, array signal processing and wireless communications. Nevertheless, previous research on tensor data analytics is mainly from a deterministic perspective, and thus cannot quantify the model uncertainty and control the model complexity, which is essentially important in model critique and avoidance of over-fitting. Fortunately, Bayesian modeling and inference provide a natural recipe, and has triggered a recent growth of interest in developing Bayesian structured tensor data analytics.
This talk consists of two parts. In this first part, we will give an introduction to the tensor decomposition methods used in modern big data analytics. In the second part, we will present the foundations of Bayesian tensor data analytics and introduce its recent advances, with a particular focus on tensor canonical polyadic decomposition model. (This lecture is co-listed with the course SI231b: Matrix Computations.)
Dr. Lei Cheng received the B.Eng. degree from Zhejiang University in 2013, and the Ph.D. degree from the University of Hong Kong in 2018. Currently, he is a Research Scientist II (Research Associate) in Shenzhen Research Institute of Big Data, The Chinese University of Hong Kong, Shenzhen. His research interests are in tensor data analytics, statistical inference and large-scale optimization.