Hyunjin Seo
I'm a Ph.D. student at KAIST, advised by Prof. Sungsoo Ahn.
My research primarily focuses on geometric deep learning, specifically within two main areas: graph neural networks (GNNs) and AI4science, including applications in molecular ML and genomics.
Initially, my research was centered on enhancing graph reasoning by integrating large language models (LLMs) with GNNs.
I've also contributed practically by developing pruning and calibration techniques for GNNs to improve computational efficiency and reliability.
Recently, I've expanded into the AI4science domain, developing an energy-based graph transformer for predicting ground-state molecular conformations.
Additionally, my latest work involves applying masked diffusion models effectively in molecule generation by orchestrating molecular component-wise forward diffusion trajectories.
