Yu (Benjamin) Pan
Assistant Professor at the University of Nebraska-Lincoln
Research: LLM Tool Using | AIOS | AIDB | Graph Learning | Manifold Learning | Representation Learning | HPC
Assistant Professor at the University of Nebraska-Lincoln
Research: LLM Tool Using | AIOS | AIDB | Graph Learning | Manifold Learning | Representation Learning | HPC
Yu Pan is an assistant professor at the University of Nebraska-Lincoln. He obtained a BS in Computer Science and Technology from University of Electronic Science and Technology of China (UESTC), an MS in Information Management from Illinois Institute of Technology (IIT) and a PhD in Computer Science from University of Nebraska-Lincoln (UNL).
His current research interest focuses on Large Language Model tool learning, AIOS and AIDB, trying to integrate large language models (LLMs) into the cloud based data infrastructure. By utilizing the reasoning capability of LLMs, the new layer of AIOS decomposes user's high level instruction into low level commands, which will liberate users from mastering detailed commands of the OS or database.
He is responsible for leading the development of the Scientific Data Management Infrastructure for the Institute of Agriculture and Natural Resources (IANR) of UNL. This data management infrastructure implements the principle of Findable, Accessible, Interoperable and Reusable (FAIR) by standardizing the data management pipeline and spatiotemporally co-locating the heterogeneous datasets from across disciplines and formats. The aim of this system is to facilitate scientific data management and analysis for the researchers from IANR and hopefully a more general community from other research domains.
His research also focus on Graph Learning and Manifold Learning, trying to gain key insight and meaningful representation from datasets defined on a graph or Euclidean space. Now collaborating with domain scientists from Biological System Engineer and Meteorology, he helps the researchers from interdisciplinary domains to gain a deeper understanding of their datasets by transforming raw data to a more meaningful representation with deep learning based methods, and by visualizing the meaningful representation with various front-end technologies. To reach this goal, research is conducted on data indexing and partitioning strategies and high performance computing frameworks such as MPI and CUDA is also utilized. He has published 18 papers in relevant conferences and journals.