8000 GitHub - kehlert/kelly_ucb: Optimal betting for a multi-armed bandit.
[go: up one dir, main page]
More Web Proxy on the site http://driver.im/
Skip to content

kehlert/kelly_ucb

Repository files navigation

If we are betting on a slot machine, then the natural question to ask is “how much should we bet”? If the probability of winning is p, it turns out that betting a fraction of our wealth equal to 2p − 1 is an optimal strategy in many senses. However, if we are faced with many slot machines with unknown probabilities of winning, then how should we bet, and which slot machines should we bet on? We will address this modified multi-armed bandit problem.

The writeup is at "./paper/kelly_ucb.pdf". This was a project for Prof. Robert Nowak's machine learning course.

About

Optimal betting for a multi-armed bandit.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published
0