Shuangfei Fan, DAC Ph.D. student in the Department of Computer Science

Graphic is from Fan’s research on “Deep Generative Models for Generating Labeled Graphs”

In disease control and prevention, understanding how an emerging infectious disease can spread beyond the visible network is important.

Marketers posting ads on an online social network can benefit from knowing how their information will spread beyond ego networks.

These two scenarios provide good examples for practical application of Shuangfei Fan’s research using deep representation learning algorithms on labeled graphs to model graph generation and graph evolution.

“A model of graph evolution would be a powerful tool for both predicting the future and the transformation of networks,” said Fan, a Discovery Analytics Center Ph.D. student in computer science.

Fan said that graphs are complex and versatile data structures that can be used to represent various kinds of real-world data with complex relationships. However, some special properties of graphs — such as discrete form and order-invariance — make generation of graphs a harder problem than it might be for other data types such as images and natural language.

“So this is a challenging and interesting area to explore,” said Fan, who earned a bachelor’s degree in computer science and technology from the University of Electronic Science and Technology of China.

At a workshop on Deep Generative Models for Highly Structured Data at the 2019 International Conference on Learning Representations, Fan presented the work she collaborated on with her advisor Bert Huang, “Deep Generative Models for Generating Labeled Graphs.”

Her other work with Huang includes “Recurrent Collective Classification. Knowledge and Information System,” in 2018, and “Training Iterative Collective Classifiers with Back-Propagation,” presented at the 12th Workshop on Mining and Learning with Graphs in August 2016.

Fan met Huang after she had already began her Ph.D. program at Virginia Tech. “I asked to join his Machine Learning Laboratory because we had the same research interests and luckily he said ‘yes,’” Fan said.

“At the Discovery Analytics Center I have had the opportunity to work with many talented and great researchers and to access computing resources that are valuable to my work,” she said.

Fan will be graduating at the end of the year and is planning on an industry career.