Hangfeng He

Hangfeng He joined the faculty at the University of Rochester after completing a PhD in computer and information science at the University of Pennsylvania.

He’s research aims to create intelligent agents to assist humans, so that people and computers can communicate better. By designing suitable formulations for tasks that require complex reasoning as well as analyzing the structure of data in different media formats, He hopes to produce simple, efficient models for real-world applications. In a paper in Proceedings of the National Academy of Sciences, He and his colleagues proposed the Layer-Peeled Model, a simple yet effective modeling strategy for understanding the deep neural networks used to develop virtual assistants, autonomous vehicles, product recommendations, and other applications.

He uses machine learning and natural language processing to further understand and analyze the structure of real-world data while simplifying modeling without the need for large-scale data.