#  Cautious Bandits 

 



####  calendar\_today Date and Time 

 **October 15, 2021** 

 01:00PM - 02:30PM EDT 

 



 

 **Presenter**: Andy Haupt  
**Topic**: Cautious Bandits

##  **Fall 2021**

 The EconCS Group holds an Economics and Computer Science research seminar each semester.

 Fall 2021 meetings are held at 1 - 2:30 PM on Fridays. Seminar Coordinators for Fall '21 are [Srivatsa R Sai](mailto:saisr@g.harvard.edu) and [Daniel Halpern](mailto:dhalpern@g.harvard.edu).

 **Abstract:** We introduce and characterize revealed risk preferences of bandit algorithms. An algorithm for the stochastic bandit problem is risk averse if for any fixed noise levels and time, there is a reward difference such that the algorithm chooses a less risky arm over a higher expected reward risky arm, with high probability in time. We experimentally find that several classical adversarial and stochastic bandit algorithms (eps-Greedy, UCB, EXP3) and prove that eps-Greedy is risk-averse. We discuss implications for the separation of learning and deployment of reinforcement learning algorithms and discuss extensions of our statement to mean-based bandit algorithms (Braverman et al. 2018) and to multi-agent environments.



 

 



 

 See also:- [ Fall 2021 EconCS Seminars ](/taxonomycalendarseminar/seminars-2021-fall)
 
 

 Share on:- [     Facebook ](#)
- [     Twitter ](#)
- [     Linkedin ](#)
 


 Save: [ Add to calendar calendar\_today ](https://econcs.seas.harvard.edu/node/1618795/event-feed.ics)  Copy link link