Hyosoon Jang
Hyosoon Jang

Hyosoon Jang

Ph.D. Student · Graduate School of AI, KAIST

Email Google Scholar GitHub LinkedIn

About

I'm Hyosoon Jang, a second-year Ph.D. student advised by Sungsoo Ahn at KAIST. My research focuses on drug discovery, with recent work on foundation models for protein and molecular learning P2, P3, and earlier work on generative modeling, including large language models P1, C4, generative flow networks C2, C3, diffusion probabilistic models C1, and variational autoencoders J2. I am also interested in machine learning for medical domains J1.

Since October 2025 I have been part of the Korean AI-specialized foundation model project (특파모), K-Fold, supported by a large-scale computing environment of 256× NVIDIA B200 GPUs. There I lead the development of one of the main components: protein representation learning for protein co-folding P3.

Publications

‡ equal contribution · * corresponding author

Synthesizing State-of-the-Art Structure Predictions from Soup of Co-folding Models
On-going

Synthesizing State-of-the-Art Structure Predictions from Soup of Co-folding Models

Hyosoon Jang, Taewon Kim, Sungsoo Ahn

Paper
Atom-level Protein Representation Learning Improves Protein Structure Prediction
P3 arXiv 2026

Atom-level Protein Representation Learning Improves Protein Structure Prediction

Hyosoon Jang‡, Taewon Kim‡, Hyunjin Seo, Seonghwan Seo, Hyeongwoo Kim, Wonho Zhung, Mingyeong Shin, Wooyoun Kim, Sungsoo Ahn

Paper arXiv Code Project Models
A Systematic Evaluation of Co-folding Model Representations for Small-Molecule Learning
P2 arXiv 2026

A Systematic Evaluation of Co-folding Model Representations for Small-Molecule Learning

Hyosoon Jang, Hyunjin Seo, Honghui Kim, Seonghyun Park, Taewon Kim, Yunhui Jang, Sungsoo Ahn

Paper arXiv Code
Self-Training Large Language Models with Confident Reasoning
C4 EMNLP 2025

Self-Training Large Language Models with Confident Reasoning

Hyosoon Jang, Yunhui Jang, Sungjae Lee, Jungseul Ok, Sungsoo Ahn

Paper PDF Code
Can LLMs Generate Diverse Molecules? Towards Alignment with Structural Diversity
P1 arXiv 2025

Can LLMs Generate Diverse Molecules? Towards Alignment with Structural Diversity

Hyosoon Jang, Yunhui Jang, Jaehyung Kim, Sungsoo Ahn

Paper arXiv
Pessimistic Backward Policy for GFlowNets
C3 NeurIPS 2024

Pessimistic Backward Policy for GFlowNets

Hyosoon Jang, Yunhui Jang, Minsu Kim, Jinkyoo Park, Sungsoo Ahn

Paper Code
Learning Energy Decomposition for Partial Inference in GFlowNets
C2 ICLR 2024 · Oral, 85/7262 = 1.16%

Learning Energy Decomposition for Partial Inference in GFlowNets

Hyosoon Jang, Minsu Kim, Sungsoo Ahn

Paper Code
Diffusion Probabilistic Models for Structured Node Classification
C1 NeurIPS 2023

Diffusion Probabilistic Models for Structured Node Classification

Hyosoon Jang, Seonghyun Park, Sangwoo Mo, Sungsoo Ahn

Paper Code
De novo Drug Design through Gradient-based Regularized Search in Information-theoretically Controlled Latent Space
J2 JCAMD 2024

De novo Drug Design through Gradient-based Regularized Search in Information-theoretically Controlled Latent Space

Hyosoon Jang, Sangmin Seo, Sanghyun Park, Byung Ju Kim, Geon-Woo Choi, Jonghwan Choi, Chihyun Park

Paper
Machine Learning Algorithms Using Systemic Inflammatory Markers for Predicting the Oncological Outcomes of Colorectal Cancer after Surgery
J1 Ann. Surg. Oncol. 2023

Machine Learning Algorithms Using Systemic Inflammatory Markers for Predicting the Oncological Outcomes of Colorectal Cancer after Surgery

Songsoo Yang‡, Hyosoon Jang‡, Inkyu Park‡, Sunhye Lee, Gaeul Oh, Chihyun Park*, Jeonghyun Kang*

Paper