Yeseul Jeon

Employment 

• Research Fellow, Department of Statistics, Texas A&M University, and  Department of Epidemiology & Biostatistics, University of California, San Francisco,  September 2024 -

• Visiting Scholar, Department of Bioinformatics, The Ohio State University, June 2022 - July 2023

• AI Researcher, Laboratory for Artificial Intelligence, Vive Company, February 2016 - February 2021

Education

• Ph.D. in Statistics, Yonsei University, 2024 (Thesis advisors: Dr. Jaewoo Park co-advised by Dr. Ick-Hoon Jin)

• B.A. in Statistics, Sungshin Women's University, 2016

Awards

• Korean International Statistical Society (KISS) - Outstanding Student Paper Award, January 2023.

• Southern Regional Council on Statistics (SRCOS)- Student Travel Award, October 2022. 

• Joint Statistical Meetings (JSM) - Winning the Text Analysis Interest Group Presentation Competition Interest Group of the American Statistical Association (ASA), August 2021. 

Research Interest

My current research focuses on developing explainable Bayesian deep learning models that seamlessly integrate the interpretability of statistical frameworks with the representational power of deep learning. These methods aim to enhance both predictive performance and interpretability while providing principled uncertainty quantification grounded in Bayesian theory. This line of research is especially meaningful in complex domains such as spatial statistics, where data are often high-dimensional and structured across space or time. By leveraging the strengths of both deep learning and Bayesian modeling, I strive to build models that are not only accurate but also trustworthy and interpretable.

During my Ph.D., my research centered on the development of methodological tools in three main areas: (i) interpretable Bayesian deep learning for high-dimensional and correlated data, (ii) latent space models for complex networks, and (iii) spatial modeling approaches for transcriptomic data. I proposed Bayesian convolutional neural network-based generalized linear models that adapt to structured input data such as images and geospatial features. I also developed novel Bayesian network models to capture latent interactions within graphs, with applications to topic modeling and psychometrics. Lastly, I designed Bayesian spatial modeling frameworks to enhance statistical inference in emerging biomedical data, such as spatial transcriptomics.

My methodological work has been applied to a wide range of scientific problems. I have substantial experience with neuroimaging data—including fMRI and histological image data—where high dimensionality and spatial dependence present unique challenges. In the biomedical domain, I have worked on spatial transcriptomics data to identify factors affecting statistical power and integrate spatial features from histological images for improved cell clustering. I have also analyzed environmental and public health data, including California wildfire data, malaria incidence, and hurricane storm surge simulations (SLOSH). These applications highlight my interest in solving real-world problems through sophisticated spatial and high-dimensional modeling techniques.

In addition to my academic research, I worked as a research scientist in an AI lab, where I analyzed diverse structured and unstructured corporate datasets. My primary focus was on natural language processing, including sentiment keyword analysis. I am currently collaborating with the Korean Fire Insurance Association, providing statistical consulting on both structured and unstructured fire-related datasets.