[Graduate Student Bok Jinseong] Awarded the Graduate Student Paper Presentation Award at the 2024 Korean Statistical Society Winter Academic Paper Presentation Conference

Paper Title: Clustered Hidden Markov Models

Paper Introduction:
In this study, we propose a new approach called the Clustered Hidden Markov Model (CHMM) to simplify the vast state space of the Hidden Markov Model (HMM). The CHMM is composed of two hidden processes and one observed process. It structures the complex state space by leveraging the clustered states of the first hidden process and the dependency between the two hidden processes. This structure is identified through maximum likelihood estimation with a Group Pairwise-difference SCAD (Smooth Clipped Absolute Deviation) penalty, and the asymptotic properties of the proposed estimator have been theoretically proven. As a result, the CHMM facilitates easier interpretation of complex data and has demonstrated excellent performance on real datasets (e.g., protein structure sequences).

Award Acceptance Speech:
It was a very meaningful experience to propose a new direction in statistical modeling—one that enhances interpretability by modifying existing models while effectively modeling complex data structures. I am deeply grateful to Professor Shin Sunyoung for her guidance that made this research possible, and I will continue to pursue meaningful research in the future.