Skip to Main Content
Frequently Asked Questions
Submit an ETD
Global Search Box
Need Help?
Keyword Search
Participating Institutions
Advanced Search
School Logo
Files
File List
Thesis.pdf (166.04 KB)
ETD Abstract Container
Abstract Header
Zooming Algorithm for Lipschitz Bandits with Linear Safety Constraints
Author Info
Hu, Tengmu
Permalink:
http://rave.ohiolink.edu/etdc/view?acc_num=osu1610134946743987
Abstract Details
Year and Degree
2021, Master of Science, Ohio State University, Actuarial and Quantitative Risk Management.
Abstract
Multi-armed bandit problems have been widely studied by many researchers over the decades. A typical multi-armed bandit problem studies algorithms for choosing from a set of actions each trial to maximize the total payoff of chosen actions. Although many learning algorithms have been developed to find an optimal strategy some challenges remain to be addressed. On one hand, when the action space is very large, an algorithm will inevitably take a long time to identify the an optimal policy. On the other hand, most of the algorithms do not differentiate between safe and unsafe actions. Although Lipschitz bandits have been used to study problems with large action spaces by assuming similarity between arms, this still does not guarantee the safety of the chosen policy. Researchers have started to look for algorithms that could not only find an optimal policy within a large set of actions but guarantee the safety of this policy. This thesis starts from a different type of Lipschitz bandit problems with safety constraints. Algorithms for Lipschitz bandit problems with safety constraint are introduced, with theoretical optimal upper bound on regret. In particular, this thesis extends the zooming algorithm to the settings where safe actions are considered.
Committee
Chunsheng Ban (Advisor)
Parinaz Naghizeh (Committee Member)
Pages
28 p.
Subject Headings
Computer Science
;
Industrial Engineering
;
Mathematics
Keywords
Reinforcement Learning
Recommended Citations
Refworks
EndNote
RIS
Mendeley
Citations
Hu, T. (2021).
Zooming Algorithm for Lipschitz Bandits with Linear Safety Constraints
[Master's thesis, Ohio State University]. OhioLINK Electronic Theses and Dissertations Center. http://rave.ohiolink.edu/etdc/view?acc_num=osu1610134946743987
APA Style (7th edition)
Hu, Tengmu.
Zooming Algorithm for Lipschitz Bandits with Linear Safety Constraints.
2021. Ohio State University, Master's thesis.
OhioLINK Electronic Theses and Dissertations Center
, http://rave.ohiolink.edu/etdc/view?acc_num=osu1610134946743987.
MLA Style (8th edition)
Hu, Tengmu. "Zooming Algorithm for Lipschitz Bandits with Linear Safety Constraints." Master's thesis, Ohio State University, 2021. http://rave.ohiolink.edu/etdc/view?acc_num=osu1610134946743987
Chicago Manual of Style (17th edition)
Abstract Footer
Document number:
osu1610134946743987
Download Count:
137
Copyright Info
© 2021, all rights reserved.
This open access ETD is published by The Ohio State University and OhioLINK.