Skip to content

This package is a possible implementation of the UCB1 selector. This selector can be used for Algorithm selection. Here it is used with random variables having different distributions.

master
Go to file
Code

Latest commit

 

Git stats

Files

Permalink
Failed to load latest commit information.
Type
Name
Latest commit message
Commit time
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

README.md

Multi-Armed Bandit selection

Description

This package is a possible implementation of the UCB1 selector. This selector can be used for Algorithm selection.

Here it is used with random variables having different distributions and from data in a CSV file.

Example

Here is an example of run with four random variables.

screen

Paper

The algorithm developed here is from:

Auer, P., Cesa-Bianchi, N., & Fischer, P. (2002). Finite-time analysis of the multiarmed bandit problem. Machine learning, 47(2-3), 235-256.

Prerequisites

This package needs Tkinter. Optional : Matplotlib

Uses

Launch the package:

python -m Bandit

Using your data files

In order to use your own csv file, you must format yur csv file as follow:

The first line contains the names of the variable/Algorithm
All the other lines contains the value of the variable/algorithm in the same column.

An example is provided here.

using any kind of files

For using this code with any kind of file, you can use your own file reader and then use the MachineList class having a list of the values as argument.

About

This package is a possible implementation of the UCB1 selector. This selector can be used for Algorithm selection. Here it is used with random variables having different distributions.

Resources

License

Releases

No releases published

Packages

No packages published

Languages

You can’t perform that action at this time.