amirbalef / PS_MOMAB

Multi-Objective Multi-Armed Bandit

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

PS_MOMAB

Description

This is an implementation of the proposed algorithms in Piecewise-Stationary Multi-Objective Multi-Armed Bandit with Application to Joint Communications and Sensing and Adaptive Energy-Efficient Waveform Design For Joint Communication and Sensing using Multiobjective Multiarmed Bandits. These algorithms are UCB-based Multi-Objective Multi-Armed Bandit and can work in non-stationary environments.

How to Install:

# In the project root folder
pip install -r requirements.txt

How to Run:

The environment can be configured in the "config_synthetic.py" file of the root folder. You then can run the experiment and save the result by commands below:

# In the project root folder
python ./run_synthetic.py
python ./plot_performances.py

Result of the synthetic dataset

Part 1

A plot of mean rewards over time. Part1

Part 2

A plot of cumulative pareto regrets over time (averaged over 100 runs each). Part2

References

About

Multi-Objective Multi-Armed Bandit


Languages

Language:Python 100.0%