Preprints

1. Nearly Optimal Active Preference Learning and Its Application to LLM Alignment

   Yao Zhao and Kwang-Sung Jun

   [arXiv], 2026
2. Beyond RLHF: A Unified Theoretical Framework of Alignment

   Jihun Yun, Juno Kim, Jongho Park, Junhyuck Kim, Jongha Jon Ryu, Jaewoong Cho, and Kwang-Sung Jun

   [arXiv], 2025

2026

1. Achieving Adaptivity and Optimality for Multi-armed Bandits using Exponential-Kullback Leibler Maillard Sampling

   Hao Qin, Kwang-Sung Jun, and Chicheng Zhang

   Transactions on Machine Learning Research (TMLR)
2. Second-Order Bounds for [0,1]-Valued Regression via Betting Loss

   Yinan Li, Sungjoon Yoon, Ethan Huang, and Kwang-Sung Jun

   In Proceedings of the Conference on Learning Theory (COLT)
3. Coverage Improvement and Fast Convergence of On-policy Preference Learning

   Juno Kim, Jihun Yun, Jason D. Lee, and Kwang-Sung Jun

   In Proceedings of the International Conference on Machine Learning (ICML)
4. Fixed Budget is No Harder Than Fixed Confidence in Best-Arm Identification up to Logarithmic Factors

   Kapilan Balagopalan, Yinan Li, Yao Zhao, Tuan Ngo Nguyen, Anton Daitche, Houssam Nassif, and Kwang-Sung Jun

   In Proceedings of the International Conference on Machine Learning (ICML)
5. Instance-Dependent Fixed-Budget Pure Exploration in Reinforcement Learning

   Yeongjong Kim, Yeoneung Kim, and Kwang-Sung Jun

   In Proceedings of the International Conference on Learning Representations (ICLR)
6. GL-LowPopArt: A Nearly Instance-Wise Minimax Estimator for (Adaptive) Generalized Linear Low-Rank Trace Regression

   Junghyun Lee, Kyoungseok Jang, Kwang-Sung Jun, Milan Vojnovic, and Se-Young Yun

   In International Conference on Artificial Intelligence and Statistics (AISTATS)

2025

1. Fixing the Loose Brake: Exponential-Tailed Stopping Time in Best Arm Identification

   Kapilan Balagopalan, Tuan Ngo Nguyen, Yao Zhao, and Kwang-Sung Jun Jun

   In Proceedings of the International Conference on Machine Learning (ICML)
2. Improved Offline Contextual Bandits with Second-Order Bounds: Betting and Freezing

   J. Jon Ryu, Jeongyeol Kwon, Benjamin Koppe, and Kwang-Sung Jun Jun

   In Proceedings of the Conference on Learning Theory (COLT)

   Video
3. HAVER: Instance-Dependent Error Bounds for Maximum Mean Estimation and Applications to Q-Learning

   Tuan Ngo Nguyen, Jay Barrett, and Kwang-Sung Jun

   In International Conference on Artificial Intelligence and Statistics (AISTATS)
4. Minimum Empirical Divergence for Sub-Gaussian Linear Bandits

   Kapilan Balagopalan and Kwang-Sung Jun

   In International Conference on Artificial Intelligence and Statistics (AISTATS)

2024

1. Adaptive Experimentation When You Can’t Experiment

   Yao Zhao, Kwang-Sung Jun, Tanner Fiez, and Lalit Jain

   In Advances in Neural Information Processing Systems (NeurIPS)
2. A Unified Confidence Sequence for Generalized Linear Models, with Applications to Bandits

   Junghyun Lee, Se-Young Yun, and Kwang-Sung Jun

   In Advances in Neural Information Processing Systems (NeurIPS)

   Oral Presentation at ICML’24 Workshop on Aligning Reinforcement Learning Experimentalists and Theorists
3. Transfer Learning in Bandits With Latent Continuity

   Hyejin Park, Seiyun Shin, Kwang-Sung Jun, and Jungseul Ok

   IEEE Transactions on Information Theory
4. Better-than-KL PAC-Bayes Bounds

   Ilja Kuzborskij, Kwang-Sung Jun, Yulian Wu, Kyoungseok Jang, and Francesco Orabona

   In Proceedings of the Conference on Learning Theory (COLT)
5. Efficient Low-Rank Matrix Estimation, Experimental Design, and Arm-Set-Dependent Low-Rank Bandits

   Kyoungseok Jang, Chicheng Zhang, and Kwang-Sung Jun

   In Proceedings of the International Conference on Machine Learning (ICML)

   Video Code
6. Noise-Adaptive Confidence Sets for Linear Bandits and Application to Bayesian Optimization

   Kwang-Sung Jun and Jungtaek Kim

   In Proceedings of the International Conference on Machine Learning (ICML)
7. Improved Regret Bounds of (Multinomial) Logistic Bandits via Regret-to-Confidence-Set Conversion

   Junghyun Lee, Se-Young Yun, and Kwang-Sung Jun

   In International Conference on Artificial Intelligence and Statistics (AISTATS)

   Code
8. Tight Concentrations and Confidence Sequences From the Regret of Universal Portfolio

   Francesco Orabona and Kwang-Sung Jun

   IEEE Transactions on Information Theory

2023

1. Kullback-Leibler Maillard Sampling for Multi-armed Bandits with Bounded Rewards

   Hao Qin, Kwang-Sung Jun, and Chicheng Zhang

   In Advances in Neural Information Processing Systems (NeurIPS)
2. Revisiting Simple Regret: Fast Rates for Returning a Good Arm

   Yao Zhao, Connor Stephens, Csaba Szepesvári, and Kwang-Sung Jun

   Proceedings of the International Conference on Machine Learning (ICML)
3. Tighter PAC-Bayes bounds through coin-betting

   Kyoungseok Jang, Kwang-Sung Jun, Ilja Kuzborskij, and Francesco Orabona

   In Proceedings of the Conference on Learning Theory (COLT)

   Oral Presentation at ICML’23 Workshop on PAC-Bayes Meets Interactive Learning

2022

1. PopArt: Efficient Sparse Regression and Experimental Design for Optimal Sparse Linear Bandits

   Kyoungseok Jang, Chicheng Zhang, and Kwang-Sung Jun

   In Advances in Neural Information Processing Systems (NeurIPS)
2. Improved regret analysis for variance-adaptive linear bandits and horizon-free linear mixture mdps

   Yeoneung Kim, Insoon Yang, and Kwang-Sung Jun

   In Advances in Neural Information Processing Systems (NeurIPS)
3. Jointly Efficient and Optimal Algorithms for Logistic Bandits

   Louis Faury, Marc Abeille, Kwang-Sung Jun, and Clément Calauzènes

   International Conference on Artificial Intelligence and Statistics (AISTATS)
4. Norm-Agnostic Linear Bandits

   Spencer Brady Gales, Sunder Sethuraman, and Kwang-Sung Jun

   In International Conference on Artificial Intelligence and Statistics (AISTATS)
5. Maillard Sampling: Boltzmann Exploration Done Optimally

   Jie Bian and Kwang-Sung Jun

   International Conference on Artificial Intelligence and Statistics (AISTATS)

   Video
6. An Experimental Design Approach for Regret Minimization in Logistic Bandits

   Blake Mason, Kwang-Sung Jun, and Lalit Jain

   Proceedings of the AAAI Conference on Artificial Intelligence (AAAI)

2021

1. Improved Confidence Bounds for the Linear Logistic Model and Applications to Linear Bandits

   Kwang-Sung Jun, Lalit Jain, Blake Mason, and Houssam Nassif

   Proceedings of the International Conference on Machine Learning (ICML)
2. Improved Regret Bounds of Bilinear Bandits using Action Space Dimension Analysis

   Kyoungseok Jang, Kwang-Sung Jun, Se Young Yun, and Wanmo Kang

   In Proceedings of the International Conference on Machine Learning (ICML)
3. Transfer Learning in Bandits with Latent Continuity

   Hyejin Park, Seiyun Shin, Kwang-Sung Jun, and Jungseul Ok

   In IEEE International Symposium on Information Theory (ISIT)

2020

1. Crush optimism with pessimism: Structured bandits beyond asymptotic optimality

   Kwang-Sung Jun and Chicheng Zhang

   Advances in Neural Information Processing Systems (NeurIPS)

   Video Slides

2019

1. Kernel Truncated Randomized Ridge Regression: Optimal Rates and Low Noise Acceleration

   Kwang-Sung Jun, Ashok Cutkosky, and Francesco Orabona

   In Advances in Neural Information Processing Systems (NeurIPS)

   Poster Slides
2. Parameter-Free Online Convex Optimization with Sub-Exponential Noise

   Kwang-Sung Jun and Francesco Orabona

   In Proceedings of the Conference on Learning Theory (COLT)
3. Parameter-Free Locally Differentially Private Stochastic Subgradient Descent

   Kwang-Sung Jun and Francesco Orabona

   In NeurIPS Workshop on Privacy in Machine Learning (PriML)

   Poster
4. Bilinear Bandits with Low-rank Structure

   Kwang-Sung Jun, Rebecca Willett, Stephen Wright, and Robert Nowak

   In Proceedings of the International Conference on Machine Learning (ICML)

   Code

2018

1. Adversarial attacks on stochastic bandits

   Kwang-Sung Jun, Y. Ma, L. Li, and Xiaojin Zhu

   In Advances in Neural Information Processing Systems (NeurIPS)
2. Data Poisoning Attacks in Contextual Bandits

   Yuzhe Ma, Kwang-Sung Jun, Lihong Li, and Xiaojin Zhu

   In Conference on Decision and Game Theory for Security (GameSec)
3. Bayesian Active Learning on Graphs

   Kwang-Sung Jun and Robert Nowak

   In Cooperative and Graph Signal Processing

2017

1. Online learning for changing environments using coin betting

   Kwang-Sung Jun, Francesco Orabona, Stephen Wright, and Rebecca Willett

   Electronic Journal of Statistics (EJS)
2. Improved Strongly Adaptive Online Learning using Coin Betting

   Kwang-Sung Jun, Francesco Orabona, Stephen Wright, and Rebecca Willett

   In Proceedings of the International Conference on Artificial Intelligence and Statistics (AISTATS), Oral Presentation
3. Scalable Generalized Linear Bandits: Online Computation and Hashing

   Kwang-Sung Jun, Aniruddha Bhargava, Robert Nowak, and Rebecca Willett

   In Advances in Neural Information Processing Systems (NeurIPS)

   Code
4. Identifying Multiple Authors in a Binary Program

   Xiaozhu Meng, Barton P. Miller, and Kwang-Sung Jun

   In Computer Security – ESORICS 2017

2016

1. Graph-based active learning: A new look at expected error minimization

   Kwang-Sung Jun and Robert Nowak

   In IEEE Global Conference on Signal and Information Processing (GlobalSIP) Symposium on Non-Commutative Theory and Applications
2. U-INVITE: Estimating Individual Semantic Networks from Fluency Data.

   Jeffrey C Zemla, Yoed N Kenett, Kwang-Sung Jun, and Joseph L Austerweil

   In Proceedings of the Annual Meeting of the Cognitive Science Society (CogSci)
3. Anytime exploration for multi-armed bandits using confidence information

   Kwang-Sung Jun and Robert Nowak

   In Proceedings of the International Conference on Machine Learning (ICML)

   post-publication note

   Emilie Kaufmann and her group have pointed out an error in this paper. Specifically, the paper’s Lemma 7 and 8 is likely to be false. In an attempt to fix this, I realized that the stopping time’s tail probability should be controlled exponentially rather than polynomially. This led to our ICML’25 “Fixing the Loose Breke” paper that guarantees an exponential tail. We have not tried hard to see if this helps resolve the issue we had in the original paper, though.

4. Top arm identification in multi-armed bandits with batch arm pulls

   Kwang-Sung Jun, Kevin Jamieson, Robert Nowak, and Xiaojin Zhu

   In International Conference on Artificial Intelligence and Statistics (AISTATS)

2015

1. Human memory search as initial-visit emitting random walk

   Kwang-Sung Jun, Xiaojin Zhu, T. Rogers, Z. Yang, and M. Yuan

   In Advances in Neural Information Processing Systems (NeurIPS)

2013

1. Learning from human-generated lists

   Kwang-Sung Jun, Xiaojin Zhu, B. Settles, and Timothy T. Rogers

   In Proceedings of the International Conference on Machine Learning (ICML)

   Video Code

2012

1. Learning from Bullying Traces in Social Media

   Jun-Ming Xu, Kwang-Sung Jun, Xiaojin Zhu, and Amy Bellmore

   In Proceedings of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (NAACL HLT)

2010

1. Cognitive models of test-item effects in human category learning

   Xiaojin Zhu, B.R. Gibson, Kwang-Sung Jun, Timothy T. Rogers, J. Harrison, and C. Kalish

   In Proceedings of the International Conference on Machine Learning (ICML)