A Lecture Entitled "Deep Learning and Scientific Computing" by Prof. Jinchao Xu from the Pennsylvania State University

Guest Introduction

Professor Xu, a Verne M. Willaman professor at The Pennsylvania State University, director of the PSU-PKU Joint Research Center for Computational Mathematics and Applications, has been authorized the position of professor fellowship of Changjiang Scholar Program of Peking University and Distinguished Youth (Class B) before. Professor Xu won the first Fung Kang Scientific Computing Award in 1995, the German ‘Humboldt’ Senior Scientist Award in 2005, the China Outstanding Youth Fund (Type B) in 2006, and the 6th International Industrial and Applied Mathematics in 2007, he was also invited to give a special report at the conference. He was an invited speaker at the International Congress of Mathematicians (2010). He was elected as a Fellow of the American Society of Industrial and Applied Mathematics (2011) and a Fellow of the American Mathematical Society (2011).


Prof. Jinchao Xu

Lecture Detail

"Scientific Computing in Data Science has been considered a vital approach to confront different types of problems amid the growing popularity of Artificial Intelligence with algorithms highly connected to mathematical methodologies. Valuable skills such as logistic regression, support vector machine, deep neural networks and convolutional neural networks, which were discussed specifically and in detail by Professor Xu in this lecture, demonstrated the close relation between mathematical algorithms and machine learning.


In the lecture

A very fundamental component in the field of machine learning is Classification. Further, Professor Xu indicated that Optimization is essential to construct a proper division or categorization of a simplified system, by using either linear or nonlinear models. Regarding different methods in classification that are linearly separable, Professor Xu pointed out that Logistic Regression and Support Vector Machine are particularly useful. For nonlinear situation, however, the problem is more complex and requires more efficient solutions which are combinations of various multigrid methods with Convolutional Neural Networks based on the Gradient Descent method, known as MgNet. The extraordinary reduction in computing time demonstrates that MgNet has amazing performance on fast training algorithms and improving generalization accuracy.

For illustrative purpose, Professor Xu discussed an application about the detection of pregnancy by pulse examination, which was popular in traditional Chinese medical science. Finally, Professor Xu concluded his speech with his vision of the unlimited research and application opportunities in the area of data science, encouraging researchers devote more time and effort in this fruitful area".

Q&A Session

In this Q&A Session, students and professors asked questions regarding the relationship between statistics and data science and Prof. Xu gave a detailed answer one by one. And discussion was actively engaged.


Active discussion


A student in the audience is asking question

Finally, to express our gratitude, Prof. Shao presented the honorary certificate to Prof. Xu  on behalf of College of Science.


Prof. Qiman Shao and Prof. Jinchao Xu