Seokhun Park

I am a Ph.D. candidate in the Department of Statistics at Seoul National University (SNU). My research interests are Interpretable AI, Functional ANOVA model, Bayesian machine learning and Quantization.

Recently, my research has focused on bridging statistical rigor with modern deep learning architectures. I am particularly interested in the interpretability of Vision Transformers (ViT) and in developing Concept Bottleneck Models (CBM) to create more transparent and trustworthy AI systems. In addition, I am interested in understanding large language models (LLMs) through Sparse Autoencoders (SAEs), with the goal of uncovering structured and interpretable representations within their internal activations. I am also interested in model quantization, particularly in improving the efficiency of deep learning models while preserving performance and interpretability. 

 • Research area

Interpretable AI, Bayesian machine learning, Quantization, Statistical learning theory

• Education

03/2020 - Present : Ph.D candidate in Statistics, Seoul National University (Advisor : Prof. Yongdai Kim) 

03/2013 - 09/2019: B.S. in Applied Mathematics, Kyung Hee University