Title: Interpretable machine learning based on visual analysis

Speaker: Shixia Liu

Time: Tuesday, November 9, 2019 14:00-15:30

Venue: 601 Conference Room, Executive Building, Jilin University Central Campus

Organizer: School of Artificial Intelligence

Abstract

Interpretable machine learning aims to make the decision-making process of the machine learning model more transparent to researchers and practitioners, thus enabling effective communication and collaboration between humans and computation systems. This report will introduce our proposed visual analysis framework for machine learning models. Our framework jumps out of the traditional "analysis then visualization" single direction visual analysis mechanism by organically combines machine learning methods and interactive visualization methods. It can better help users understand and analyze complex models and their output, thus constantly improve the machine learning model. We provide the technical basis for users to select, utilize and improve machine learning models. Finally, combined with specific application examples, such as deep learning model and integrated learning model analysis, we introduce our visual analysis technology based on the framework.

Short biography of Dr. Shixia Liu

I am an associate professor in the School of Software, Tsinghua University. I received a B.S. and M.S. in Computational Mathematics from Harbin Institute of Technology, a Ph.D. in Computer Aided Design and Computer Graphics from Tsinghua University. Before I joined Tsinghua, I worked as a lead researcher at Microsoft Research Asia and a research staff member and research manager at IBM China Research Lab. My research tightly integrates interactive visualization with machine learning or data mining techniques to help users consume huge amounts of information. Specifically, my research interests include visual text analytics, visual social media analytics, and text mining.