StrategicCUCB

The repo features the experimental analysis on Combinatorial Bandits under Strategic Manipulations.

Dependency

• python3.6.5
• numpy
• matplotlib

Experiment

• For testing Combinatorial UCB and its strategic variant, please modify the code eval.py accordingly, then

python3 eval.py

• For testing optimal collusion strategy, please modify the code strategy.py accordingly, then

python3 strategy.py

Contributions and Contact

This repo is built on code from (https://github.com/lilianweng/multi-armed-bandit).
The experiment is mainly done by Ke Li and some parts done by Jing Dong.
Please contact keli@link.cuhk.edu.cn or jindong@umich.edu if you have any question regarding this repo.