Mengjia Xu, PhD
Postdoctoral Associate, MIT McGovern Institute and MGH
In real world applications, networks can be used as a powerful language for describing and modeling complex systems, e.g. social networks, economic networks, biomedical networks, internet networks, etc. In the past few decades, machine learning based graph representation has become an active research area since it can provide insights into how to make good use of the information hidden in graphs and facilitate more accurate and efficient downstream graph analysis tasks (e.g., community detection or link prediction). I will first give an overview of the latest graph embedding methods, and then I will specifically introduce the principles of Gaussian embedding technique with uncertainty quantification. Finally, I will illustrate a real-world functional brain network application using the Gaussian embedding method.
Mengjia Xu is currently a postdoctoral associate at the MIT McGovern Institute and Massachusetts General Hospital. Before joining MGH in 2018, she was a postdoc of Beijing International Center for Mathematical Research at Peking University. She received her PhD degree in Computer Science from Northeastern University of China in 2017. During her PhD, she worked at Brown University for two years as visiting PhD student in the Division of Applied Mathematics. Previously, she worked as a full-time software engineer intern at Neusoft for two years. Her main research interests are to develop data-driven machine learning methods for medical image data analysis in diverse real-world applications.
Current MIT Sea Grant projects in machine learning and modeling: