Junsu Kim

Ph.D. Student, KAIST

Graduate School of Artificial Intelligence

junsu.kim [AT] kaist.ac.kr

About

I am a Ph.D. student at ALIN Lab advised by Prof. Jinwoo Shin and Prof. Kimin Lee at Korea Advanced Institute of Science and Technology (KAIST), and a visiting Ph.D student at UC Berkeley, advised by Prof. Sergey Levine. Previously, I received B.S in Electrical Engineering from KAIST in Feb. 2020. My main research question is to solve challenging decision making problems via RL (e.g., long-horizon tasks, robotic manipulation). Relevant topics include hierarchical RL and representation learning for RL. I am also broadly interested in diverse areas of reinforcement learning, such as goal-conditioned RL, model-based RL, and multi-agent RL. Prior to that, I worked on machine learning for drug discovery.

Topics of interest

Publications

(*: Equal contribution, C: Conferences, W: Workshop, P: preprint).

[C6] Multi-View Masked World Models for Visual Robotic Manipulation

Younggyo Seo*, Junsu Kim*, Stephen James, Kimin Lee, Jinwoo Shin, Pieter Abbeel

International Conference on Machine Learning (ICML), 2023

RSS Workshop on Experiment-oriented Locomotion and Manipulation Research, 2023 (Spotlight Presentation)

Paper Code Website

[W2] Fragment-based Multi-view Molecular Contrastive Learning

Seojin Kim*, Jaehyun Nam*, Junsu Kim, Hankook Lee, Sungsoo Ahn, Jinwoo Shin

ICLR Workshop on Machine Learning for Materials, 2023

Paper

[C5] Imitating Graph-Based Planning with Goal-Conditioned Policies

Junsu Kim, Younggyo Seo, Sungsoo Ahn, Kyunghwan Son, Jinwoo Shin

International Conference on Learning Representations (ICLR), 2023

Paper Code

[W1] Dynamics-Augmented Decision Transformer for Offline Dynamics Generalization

Changyeon Kim*, Junsu Kim*, Younggyo Seo, Kimin Lee, Honglak Lee, Jinwoo Shin

NeurIPS Workshop on Offline Reinforcement Learning, 2022

Paper

[C4] Disentangling Sources of Risk for Distributional Multi-Agent Reinforcement Learning

Kyunghwan Son, Junsu Kim, Sungsoo Ahn, Roben Delos Reyes, Yung Yi, Jinwoo Shin

International Conference on Machine Learning (ICML), 2022

Paper

[C3] Landmark-Guided Subgoal Generation in Hierarchical Reinforcement Learning

Junsu Kim, Younggyo Seo, Jinwoo Shin

Neural Information Processing Systems (NeurIPS), 2021

Paper Slides Poster Code

[C2] Self-Improved Retrosynthetic Planning

Junsu Kim, Sungsoo Ahn, Hankook Lee, Jinwoo Shin

International Conference on Machine Learning (ICML), 2021

Paper Slides Poster Code

[C1] Guiding Deep Molecular Optimization with Genetic Exploration

Sungsoo Ahn, Junsu Kim, Hankook Lee, Jinwoo Shin

Neural Information Processing Systems (NeurIPS), 2020

Paper Code

Education

Korea Advanced Institute of Science and Technology (KAIST)Mar. 2020 - present

M.S./Ph.D. in Artificial Intelligence

Korea Advanced Institute of Science and Technology (KAIST)Mar. 2015 - Feb. 2020

B.S. in Electrical Engineering (Summa Cum Laude)

Honors & Awards

Finalist, Qualcomm Innovation Fellowship Korea

Qualcomm, 2022

NAVER Best Paper Award

Korean Artificial Intelligence Association, 2021

Finalist, Qualcomm Innovation Fellowship Korea

Qualcomm, 2021

KT Prize, Post Corona AI Challenge: Infectious Disease Modelling

Korea Ministry of Science and ICT, National IT Industry Promotion Agency (NIPA), and KT, 2020

Qualcomm-KAIST Innovation Award

Qualcomm and KAIST, 2018

National Science and Engineering Undergraduate Scholarship

Korea Ministry of Science and ICT, 2017-2018

Invited Talks

Combining Handcrafted Search Algorithms with Deep Reinforcement Learning

Pohang University of Science and Technology (POSTECH), Jun. 2022

Academic Services

    Conference Reviewer

  • Neural Information Processing Systems (NeurIPS)
  • International Conference on Machine Learning (ICML)
  • International Conference on Learning Representations (ICLR)

Vitæ

Full CV in PDF (Last update: Aug 2023).

Junsu Kim
based on Martin Saveski's template