Professor Li Guodong of the University of Hong Kong was invited by the School of Statistics to lecture on An efficient tensor regression for high-dimensional data online. The lecture was presided by Liu Xiaohui, Professor of the School of Statistics. Some teachers and graduate students of the School of Statistics attended the lecture.

Professor Li Guodong introduced the core content of the lecture from five parts, including Motivations, Tensor train decomposition and tensor regression, Tensor train regression for high-dimensional data, Simulation experiments, Real analysis.

To begin with, Professor Li introduced the Tensor data and Matricization, which is a technique to deal with higher-order tensors. One-mode matricization and Sequential matricization are two most commonly used methods of matricization for tensor data.

Then, he explained the CP decomposition and Tucker decomposition respectively in detail.

Finally, he introduced the Tensor train decomposition and methods to apply it to tensor regression.

Following the lecture was a heated question and answer session, in which teachers and students asked questions of relevant academic issues actively. Professor Li’s wonderful lecture of An efficient tensor regression for high-dimensional data has brought professional benefits for teachers and students’ future research.

(By Yin Ying, Zhang Mengyun)