Youngsoo Jang
Research Scientist at LG AI Research, Advanced ML Lab
Research Interest: Reinforcement Learning (RL), Large Language Models (LLMs), Reinforcement Learning from Human Feedback (RLHF)
Contact: jys5609 at gmail.com
Education
2018.03 - 2022.08: Ph.D. in Computer Science, KAIST, Republic of Korea (Advisor: Prof. Kee-Eung Kim)
2016.03 - 2018.02: M.S. in Computer Science, KAIST, Republic of Korea (Advisor: Prof. Kee-Eung Kim)
2011.02 - 2016.02: B.S. in Mathematical Science and Computer Science (Double Major), KAIST, Republic of Korea
Work Experience
2022.08 - present: Research Scientist (& RL Squad Leader) at LG AI Research, Advanced ML Lab
Publications
Prospector: Improving LLM Agents with Self-Asking and Trajectory Ranking
Byoungjip Kim, Youngsoo Jang, Lajanugen Logeswaran, Geon-Hyeong Kim, Yu Jin Kim, Honglak Lee, and Moontae Lee
Conference on Empirical Methods in Natural Language Processing (EMNLP) Findings. 2024
Proceedings of Neural Information Processing Systems (NeurIPS) Foundation Models for Decision Making Workshop. 2023
Semantic Skill Grounding for Embodied Instruction-Following in Cross-Domain Environments
Sangwoo Shin*, SeungHyun Kim*, Youngsoo Jang, Moontae Lee, and Honguk Woo (*: equal contribution)
Association for Computational Linguistics (ACL) Findings. 2024
Degeneration-free Policy Optimization: RL Fine-Tuning for Language Models without Degeneration
Youngsoo Jang, Geon-Hyeong Kim, Byoungjip Kim, Yu Jin Kim, Honglak Lee, and Moontae Lee
Proceedings of International Conference on Machine Learning (ICML). 2024
International Conference on Learning Representations (ICLR) Workshop on Generative Models for Decision Making. 2024
Show, Think, and Tell: Thought-Augmented Fine-Tuning of Large Language Models for Video Captioning
Byoungjip Kim, Dasol Hwang, Sungjun Cho, Youngsoo Jang, Honglak Lee, Moontae Lee
The IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshop on Multi-Modal Foundation Models. 2024
SafeDICE: Offline Safe Imitation Learning with Non-Preferred Demonstrations
Youngsoo Jang, Geon-Hyeong Kim, Jongmin Lee, Sungryull Sohn, Byoungjip Kim, Honglak Lee, and Moontae Lee
Proceedings of Neural Information Processing Systems (NeurIPS). 2023
Information-Theoretic State Space Model for Multi-View Reinforcement Learning
HyeongJoo Hwang, Seokin Seo, Youngsoo Jang, Sungyoon Kim, Geon-Hyeong Kim, Seunghoon Hong, and Kee-Eung Kim
Proceedings of International Conference on Machine Learning (ICML). 2023
LobsDICE: Offline Imitation Learning from Observation via Stationary Distribution Correction Estimation
Geon-Hyeong Kim*, Jongmin Lee*, Youngsoo Jang, Hongseok Yang, and Kee-Eung Kim (*: equal contribution)
Proceedings of Neural Information Processing Systems (NeurIPS). 2022
GPT-Critic: Offline Reinforcement Learning for End-to-End Task-Oriented Dialogue Systems
Youngsoo Jang, Jongmin Lee, and Kee-Eung Kim
International Conference on Learning Representations (ICLR). 2022
Monte-Carlo Planning and Learning with Language Action Value Estimates
Youngsoo Jang, Seokin Seo, Jongmin Lee, and Kee-Eung Kim
International Conference on Learning Representations (ICLR). 2021
Variational Inference for Sequential Data with Future Likelihood Estimates
Geon-Hyeong Kim, Youngsoo Jang, Hongseok Yang, and Kee-Eung Kim
Proceedings of International Conference on Machine Learning (ICML). 2020
End-to-End Neural Pipeline for Goal-Oriented Dialogue System using GPT-2
Donghoon Ham*, Jeong-Gwan Lee*, Youngsoo Jang, and Kee-Eung Kim (*: equal contribution)
Proceedings of Association for Computational Linguistics (ACL). 2020
Proceedings of AAAI Conference on Artificial Intelligence (AAAI) DSTC8 Workshop. 2020
1st place on 8th Dialog System Technology Challenge (DSTC8) Multi-domain Task Completion Track, 2019
Bayes-Adaptive Monte-Carlo Planning and Learning for Goal-Oriented Dialogues
Youngsoo Jang, Jongmin Lee, and Kee-Eung Kim
Proceedings of AAAI Conference on Artificial Intelligence (AAAI). 2020
Proceedings of Neural Information Processing Systems (NeurIPS) Conversational AI Workshop. 2019 (oral presentation)
Trust Region Sequential Variational Inference
Geon-Hyeong Kim, Youngsoo Jang, Jongmin Lee, Wonseok Jeon, Hongseok Yang, and Kee-Eung Kim
Proceedings of Asian Conference on Machine Learning (ACML). 2019
PyOpenDial: A Python-based Domain-Independent Toolkit for Developing Spoken Systems with Probabilistic Rules
Youngsoo Jang*, Jongmin Lee*, Jaeyoung Park*, Kyeng-Hun Lee, Pierre Lison, and KeeEung Kim (*: equal contribution)
Proceedings of Empirical Methods in Natural Language Processing (EMNLP), System Demonstrations. 2019
Cross-language Neural Dialog State Tracker for Large Ontologies using Hierarchical Attention
Youngsoo Jang, Jiyeon Ham, Byung-Jun Lee, and Kee-Eung Kim
IEEE/ACM Transactions on Audio, Speech, and Language Processing (TASLP). 2018
Constrained Bayesian Reinforcement Learning via Approximate Linear Programming
Jongmin Lee, Youngsoo Jang, Pascal Poupart, and Kee-Eung Kim
Proceedings of International Joint Conference on Artificial Intelligence (IJCAI). 2017
ECML-PKDD Workshop on Scaling-Up Reinforcement Learning (SURL). 2017
Neural Dialog State Tracker for Large Ontologies by Attention Mechanism
Youngsoo Jang*, Jiyeon Ham*, Byung-Jun Lee, Youngjae Chang, and Kee-Eung Kim (*: equal contribution)
IEEE Workshop on Spoken Language Technology (SLT). 2016 / 3rd place on 5th Dialog State Tracking Challenge (DSTC)
Awards and Honors
NeurIPS Outstanding Reviewer Award, 2021
Qualcomm-KAIST Innovation Awards, Qualcomm, 2019 (awarded $5,000)
1st place on 8th Dialog System Technology Challenge (DSTC8) Multi-domain Task Completion Track, 2019
Naver Ph.D. Fellowship, NAVER, 2018 (awarded $5,000)
3rd place on 5th Dialog State Tracking Challenge (DSTC5), 2016
Best TA Award, Introduction to Programming (CS101, 2016 Spring), School of Computing, KAIST, Sep. 2016
Academic Talks
Degeneration-free Policy Optimization: RL Fine-Tuning for Language Models without Degeneration
2024. 07. 23. ICML 2024, Vienna, Austria
SafeDICE: Offline Safe Imitation Learning with Non-Preferred Demonstrations
2023. 12. 15. NeurIPS 2023, New Orleans, USA
GPT-Critic: Offline Reinforcement Learning for End-to-End Task-Oriented Dialogue Systems
2022. 04. 26. ICLR 2022, Virtual
2022. 05. 11. NAVER, Virtual
Monte-Carlo Planning and Learning with Language Action Value Estimates
2021. 05. 04. ICLR 2021 ML in Korea, Virtual
2021. 05. 04. ICLR 2021, Virtual
Bayes-Adaptive Monte-Carlo Planning and Learning for Goal-Oriented Dialogues
2020. 07. 13. KAKAO Brain, Pangyo, Korea
2020. 02. 11. AAAI Conference on Artificial Intelligence, New York, USA
2019. 12. 14. NeurIPS Conversational AI Workshop 2019, Vancouver, Canada
2019. 11. 29. Qualcomm-KAIST Innovation Award 2019, KAIST, Korea
2019. 09. 27. KAKAO, Jeju, Korea
Academic Services
NeurIPS Reviewer
ICML Reviewer
ICLR Reviewer
AAAI Reviewer
ACL Reviewer
EMNLP Reviewer
NAACL Reviewer