Kwang-Sung Jun – UW-Madison

profile_pic 

I am a post-doc supervised by Professors Robert Nowak, Rebecca Willett, and Stephen Wright at UW-Madison Wisconsin Institute for Discovery under optimization theme. I received my PhD in 2015 at University of Wisconsin-Madison, advised by Xiaojin (Jerry) Zhu. I obtained a B.S. in computer science with a minor in mathematics from School of Computing, Soongsil University, South Korea.

My research focuses on sequential decision-making in feedback loops (i.e., the multi-aremd bandit problem). I also work on online optimization and machine learning applied to psychology.

I am on the job market targeting to start in Fall 2018. Contact me!

Multi-Armed Bandit

Multi-armed bandit is a state-less version of the reinforcement learning (RL). However, bandits usually enjoy stronger theoretical guarantees and have abundant real-world applications. Informally speaking, bandits learn to make better decisions over time in a feedback-loop. The decisions necessarily affect the feedback information, and the feedback data collected so far is no longer i.i.d.; most traditional learning guarantees do not work.

Bandits are actively being studied in both theory and applications including deployable web service. Also, the cartoon caption contest of New Yorker is currently using a multi-armed bandit algorithm to efficiently crowdsource caption evaluations (read this article)!

Selected Publications

  1. Scalable Generalized Linear Bandits: Online Computation and Hashing.
    Kwang-Sung Jun, Aniruddha Bhargava, Robert Nowak, Rebecca Willett.
    In Neural Information Processing Systems (NIPS), 2017. [arxiv]

  2. Improved Strongly Adaptive Online Learning using Coin Betting.
    Kwang-Sung Jun, Francesco Orabona, Stephen Wright, Rebecca Willett.
    In International Conference on Artificial Intelligence and Statistics (AISTATS), 2017. Oral presentation. [official] [arxiv]

  3. Graph-Based Active Learning: A New Look at Expected Error Minimization.
    Kwang-Sung Jun and Robert Nowak.
    In IEEE GlobalSIP Symposium on Non-Commutative Theory and Applications, 2016. [ieee][arxiv]

  4. Anytime Exploration for Multi-armed Bandits using Confidence Information.
    Kwang-Sung Jun and Robert Nowak.
    In International Conference on Machine Learning (ICML), 2016. [pdf]

  5. Top arm identification in multi-armed bandits with batch arm pulls.
    Kwang-Sung Jun, Kevin Jamieson, Rob Nowak, Xiaojin Zhu.
    In International Conference on Artificial Intelligence and Statistics (AISTATS), 2016. [pdf]

  6. Human memory search as initial-visit emitting random walk.
    Kwang-Sung Jun, Xiaojin Zhu, Timothy Rogers, Zhuoran Yang, Ming Yuan.
    In Neural Information Processing Systems (NIPS), 2015. [pdf]

  7. Learning from Human-Generated Lists.
    Kwang-Sung Jun, Xiaojin Zhu, Burr Settles, Timothy Rogers.
    In International Conference on Machine Learning (ICML), 2013. [pdf] [code&data] [video]

Talks

  • 10/17: At SILO, “Scalable Generalized Linear Bandits: Online Computation and Hashing”. [abstract]

  • 04/17: At AISTATS, “Improved Strongly Adaptive Online Learning using Coin Betting”.

  • 06/16: At CPCP Annual Retreat, “Multi-Armed Bandit Algorithms and Applications to Experiment Selection”. [abstract & video]

  • 03/16: At SILO, “Top Arm Identification in Multi-Armed Bandits with Batch Arm Pulls”. [abstract & video]

  • 03/16: At Soongsil University, two talks on human memory search.

  • 06/16: At ICML, “Anytime Exploration for Multi-armed Bandits using Confidence Information”. [video]

  • 11/15: At HAMLET (interdisciplinary seminar series at UW-Madison), “Measuring semantic structure from verbal fluency data with the initial-visit-emitting (INVITE) random walk”.

  • 03/15: At TTIC, “Learning from Human-Generated Lists”.

  • 06/13: At ICML, “Learning from Human-Generated Lists”. [video]

Services

  • Program Committee / Reviewer: AISTATS’18, AAAI’18, NIPS’17, ICML’17, COLT’17 (subreviewer), AISTATS’17, ICML’16.