AI Research Scientist
LG Global AI Center, Ann Arbor, MI, USA
Contact: 24 Frank Lloyd Wright Dr., Suite A-3400, Ann Arbor, MI, 48105
Email : srsohn [dot] lgai [at] gmail [dot] com
I am an AI Research Scientist at LG Global AI Center, Ann Arbor. I received my PhD from the Computer Science Department at the University of Michigan, where I worked with Professor Honglak Lee. My research interests lie at the intersection of deep reinforcement learning and large language model (LLM), as well as in their applications to artificial intelligence. In particular, I am interested in equipping autonomous agents with the ability to perform high-level inference and reasoning, enabling them to solve complex tasks more efficiently. Specific research areas include LLM agents (i.e., large action models), meta- and multi-task reinforcement learning for task generalization, and hierarchical reinforcement learning and planning for tasks with complex structures.
[Sept 2021] I joined LG AI Global Center as a research scientist.
[July 2021] I finished my PhD at UM.
[June 2019] I will intern at Google Brain, Mountain View, advised by Jayden Ooi, Yinlam Chow and Ofir Nachuum (Summer 2019)
[June 2018] I will intern at Microsoft Research, Montreal, advised by Harm van Seijen and Mehdi Fatemi (Summer 2018)
Scalable Video-to-Dataset Generation for Cross-Platform Mobile Agents
Yunseok Jang*, Yeda Song*, Sungryull Sohn, Lajanugen Logeswaran, Tiange Luo, Dong-Ki Kim, Kyunghoon Bae, Honglak Lee
In CVPR 2025
[Coming soon]
Hierarchical Decomposition Framework for Feasibility-hard Combinatorial Optimization
Hanbum Ko*, Minu Kim*, Han-Seul Jeong, Sunghoon Hong, Deunsol Yoon, Youngjoon Park, Woohyung Lim, Honglak Lee, Moontae Lee, Kanghoon Lee, Sungbin Lim, Sungryull Sohn
In ICORES 2025 (Best student paper award)
[Coming soon]
Auto-intent: Automated intent discovery and self-exploration for large language model web agents
Jaekyeom Kim, Dong-Ki Kim, Lajanugen Logeswaran, Sungryull Sohn, Honglak Lee
In EMNLP Findings 2024
[Paper]
Autoguide: Automated generation and selection of state-aware guidelines for large language model agents
Yao Fu, Dong-Ki Kim, Jaekyeom Kim, Sungryull Sohn, Lajanugen Logeswaran, Kyunghoon Bae, Honglak Lee
In NeurIPS 2024
[Paper]
Code Models are Zero-shot Precondition Reasoners
Lajanugen Logeswaran, Sungryull Sohn, Yiwei Lyu, Anthony Zhe Liu, Dong-Ki Kim, Dongsub Shim, Moontae Lee, Honglak Lee
In NAACL 2024
[Paper]
Multiprompter: Cooperative prompt optimization with multi-agent reinforcement learning
Dong-ki Kim, Sungryull Sohn, Lajanugen Logeswaran, Dongsub Shim, Honglak Lee
ArXiv Preprint (2024)
[Preprint]
Unsupervised Object Interaction Learning with Counterfactual Dynamics Models
Jongwook Choi, Sungtae Lee, Xinyu Wang, Sungryull Sohn, Honglak Lee
In AAAI 2024
[Paper]
TOD-Flow: Modeling the Structure of Task-Oriented Dialogues
Sungryull Sohn*, Yiwei Lyu*, Anthony Zhe Liu, Lajanugen Logeswaran, Dong-ki Kim, Dongsub shim, Honglak Lee
In EMNLP 2023
[Paper]
From Heuristic to Analytic: Cognitively-Motivated Reasoning Strategies for Coherent Physical Commonsense in Pre-Trained Language Models
Zheyuan Zhang, Shane Storks, Fengyuan Hu, Sungryull Sohn, Moontae Lee, Honglak Lee, Joyce Chai
In EMNLP 2023
[Paper]
A Picture is Worth a Thousand Words: Language Models Plan from Pixels
Anthony Zhe Liu, Lajanugen Logeswaran, Sungryull Sohn, Honglak Lee
In EMNLP 2023
SafeDICE: Offline Safe Imitation Learning with Non-Preferred Demonstrations
Youngsoo Jang, Geon-Hyeong Kim, Jongmin Lee, Sungryull Sohn, Byoungjip Kim, Honglak Lee, Moontae Lee
In NeurIPS 2023
[Paper]
Unsupervised Task Graph Generation from Instructional Video Transcripts
Lajanugen Logeswaran, Sungryull Sohn, Yunseok Jang, Moontae Lee, Honglak Lee
In Findings of ACL 2023
[paper]
Fast Inference and Transfer of Compositional Task Structures for Few-shot Task Generalization
Sungryull Sohn, Hyunjae Woo, Jongwook Choi, Lyubing Qiang, Izzeddin Gur, Aleksandra Faust, Honglak Lee.
In UAI 2022 Oral presentation (5% acceptance).
Learning Parameterized Task Structure for Generalization to Unseen Entities
Anthony Z. Liu* , Sungryull Sohn*, Mahdi Qazwini, Honglak Lee.
In AAAI 2022 Oral presentation (4.5% acceptance).
[Paper][Project] [Slide][Code]
Successor Feature Landmarks for Long-Horizon Goal-Conditioned Reinforcement Learning
Chris Hoang, Sungryull Sohn, Jongwook Choi, Wilka Carvalho, Honglak Lee.
In NeurIPS 2021
Shortest-Path Constrained Reinforcement Learning for Sparse Reward Tasks
Sungryull Sohn*, Sungtae Lee*, Jongwook Choi, Harm van Seijen, Mehdi Fatemi, Honglak Lee.
In ICML 2021
Reinforcement Learning for Sparse-Reward Object-Interaction Tasks in a First-person Simulated 3D Environment
Wilka Carvalho, Anthony Liang, Kimin Lee, Sungryull Sohn, Honglak Lee, Richard L. Lewis, Satinder Singh.
In IJCAI 2021.
BRPO: Batch Residual Policy Optimization
Sungryull Sohn*, Yinlam Chow*, Jayden Ooi*, Ofir Nachum, Honglak Lee, Ed Chi, Craig Boutilier.
In IJCAI 2020.
[Paper]
Meta Reinforcement Learning with Autonomous Inference of Subtask Dependencies
Sungryull Sohn, Hyunjae Woo, Jongwook Choi, Honglak Lee.
In ICLR 2020.
Revisiting Gist-PCA Hashing for Near Duplicate Image Detection
Hyunwoo Kim, Sungryull Sohn, Junmo Kim
In Journal of Signal Processing Systems 91.6 (2019): 575-586.
Hierarchical Reinforcement Learning for Zero-shot Generalization with Subtask Dependencies
Sungryull Sohn, Junhyuk Oh, Honglak Lee
In NeurIPS 2018.
Learning to generate long-term future via hierarchical prediction
Ruben Villegas, Jimei Yang, Yuliang Zou, Sungryull Sohn, Xunyu Lin, Honglak Lee
In ICML 2017
Uncorrelated Component Analysis-Based Hashing
Sungryull Sohn, Hyunwoo Kim, Junmo Kim
In IEEE Transactions on Image Processing 26 (8), 3759-3774, 2017
Supervised Hashing via Uncorrelated Component Analysis
Sungryull Sohn, Hyunwoo Kim, Junmo Kim
In AAAI 2016
Facial age estimation via extended curvature Gabor filter
Jihwan Kim, Dongyoon Han, Sungryull Sohn, Junmo Kim
In ICIP 2015
Sep 2016 - Sep 2021
University of Michigan, Ann Arbor, MI
Ph.D. student in Computer Science and Engineering
Feb 2012 - June 2013
Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea
M.S. in Electrical Engineering
Feb 2008 - Feb 2012
Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Korea
B.S. in Electrical Engineering
LG Global AI Center, Ann Arbor, MI, USA
AI Research Scientist, Sept 2021 - Current
University of Michigan, Ann Arbor, MI, USA
Graduate Student Instructor (GSI) for EECS545 Machine Learning (instructor: Professor Honglak Lee), Jan. 2020 - Apr. 2020
Google Brain, Mountain View, CA, USA
Research Intern, June 2019 - Aug. 2019
Student Researcher, Aug. 2019 - Mar. 2020
Microsoft Research, Montreal, Canada
Research Intern, June 2018 - Sep. 2018
Electronics and Telecommunications Research Institute (ETRI), Daejeon, Korea
Researcher, Aug. 2013 - Aug. 2016
Reviewer for TPAMI, JMLR.
Reviewer for the following conferences:
ICLR 2023, ICML 2023, NeurIPS 2023
ICLR 2022, ICML 2022 (Top 10% reviewer), NeurIPS 2022
ICML 2021, NeurIPS 2021
ICML 2020 (Top 33% reviewer), NeurIPS 2020, IJCAI 2020 (Session chair)
NeurIPS 2019
Reviewer/chair for the following workshops:
OSC workshop at ICLR 2022, METALEARN workshop at NeurIPS 2022 (area chair)
L2L workshop at ICLR 2021, METALEARN workshop at NeurIPS 2021 (area chair)
L2L workshop at ICLR 2020, DRL workshop at NeurIPS 2020 (program committee), METALEARN workshop at NeurIPS 2020 (senior reviewer)
SPiRL workshop at ICLR 2019, METALEARN workshop at NeurIPS 2019
Graduate fellowship for M.S. study from KAIST, 2012–2013
Scholarship for Student Exchange Program from KAIST, 2009
National Presidential Science Scholarship for B.S. study from the Korea Student Aid Foundation, 2008–2012
Korea Mathematics Olympiad (KMO) Silver Medal, 2007
Korea Physics Olympiad (KPhO) Bronze Medal, 2007
Korea Olympiad in Informatics (KOI) Bronze Medal, 2006