Publications

Future Prediction Can be a Strong Evidence of Good History Representation in Partially Observable Environments

J. Kwon, L. Yang, R. Nowak and J. Hanna
arXiv Preprint
[Arxiv]

On the Complexity of First-Order Methods in Stochastic Bilevel Optimization

J. Kwon, D. Kwon and H. Lyu
arXiv Preprint
[Arxiv]

Prospective Side Information for Latent MDPs

J. Kwon, Y. Efroni, C. Caramanis and S. Mannor
arXiv Preprint
[Arxiv]

On Penalty Methods for Nonconvex Bilevel Optimization and First-Order Stochastic Approximation

J. Kwon, D. Kwon, S. Wright and R. Nowak
Proceedings of the 12th International Conference on Learning Representations (ICLR) 2024 (Spotlight)
[Arxiv] [Conference]

A Fully First-Order Method for Stochastic Bilevel Optimization

J. Kwon, D. Kwon, S. Wright and R. Nowak
Proceedings of the 40th International Conference on Machine Learning (ICML) 2023 (Oral Presentation)
[Arxiv] [Conference]

Feed Two Birds with One Scone: Exploiting Wild Data for Both Out-of-Distribution Generalization and Detection

H. Bai, G. Canal, X. Du, J. Kwon, R. D Nowak, Y. Li
Proceedings of the 40th International Conference on Machine Learning (ICML) 2023
[Arxiv] [Conference]

Reward-Mixing MDPs with Few Latent Contexts are Learnable

J. Kwon, Y. Efroni, C. Caramanis and S. Mannor
Proceedings of the 40th International Conference on Machine Learning (ICML) 2023
[Arxiv] [Conference]

Tractable Optimality in Episodic Latent MABs

J. Kwon, Y. Efroni, C. Caramanis and S. Mannor
Proceedings of the 36th Neural Information Processing Systems (NeurIPS) 2022
[Arxiv] [Conference]

Coordinate Attacks against Contextual Bandits: Fundamental Limits and Defense Mechanisms

J. Kwon, Y. Efroni, C. Caramanis and S. Mannor
Proceedings of the 39th International Conference on Machine Learning (ICML) 2022
[Arxiv] [Conference]

Reinforcement Learning in Reward-Mixing MDPs

J. Kwon, Y. Efroni, C. Caramanis and S. Mannor
Proceedings of the 35th Neural Information Processing Systems (NeurIPS) 2021
[Arxiv] [Conference]

MLE and EM for Well-Separated Mixtures: Minimax Rates

J. Kwon and C. Caramanis
Work in Progress
[Arxiv]

RL for Latent MDPs: Regret Guarantees and a Lower Bound

J. Kwon, Y. Efroni, C. Caramanis and S. Mannor
Proceedings of the 35th Neural Information Processing Systems (NeurIPS) 2021 (Spotlight)
[Arxiv] [Conference] [RL Theory Seminar]

On the Computational and Statistical Complexity of Over-Parameterized Matrix Sensing

J. Zhuo, J. Kwon, N. Ho and C. Caramanis
Journal of Machine Learning Research (JMLR) 2024 (To appear)
[Arxiv] [Conference]

On the Minimax Optimality of the EM Algorithm for Two-Component Mixed Linear Regression

J. Kwon, N. Ho and C. Caramanis
Proceedings of the 24th International Conference on Artificial Intelligence and Statistics (AISTATS) 2021
[Arxiv] [Conference]

The EM Algorithm gives Sample-Optimality for Learning Mixtures of Well-Separated Gaussians

J. Kwon and C. Caramanis
Proceedings of the 33rd Annual Conference on Learning Theory (COLT) 2020
[Arxiv] [Conference] [Video]

EM Converges for a Mixture of Many Linear Regressions

J. Kwon and C. Caramanis
Proceedings of the 23rd International Conference on Artificial Intelligence and Statistics (AISTATS) 2020
[Arxiv] [Conference]

Global Convergence of the EM Algorithm for Mixtures of Two Component Linear Regression

J. Kwon, W. Qian, C. Caramanis, Y. Chen, and D. Damek
Proceedings of the 32nd Annual Conference on Learning Theory (COLT) 2019
[Arxiv] [Conference] [Video]