Correlated Learning for Aggregation Systems

In the paper we propose Correlated Learning (CL) algorithm which focuses on the problem of serving individual interests, i.e., learning policies for individuals in homogeneous multi-agent environment in the presence of a self interested central entity. This repository implements traffic game example used in the paper. Traffic game is a stateless coordination game motivated by traffic control task where players need to select a route such that they do not cause congestion.

Dependencies

1. Python 3
2. tensorflow

To run the experiment $ python3 environment.py

Citing CL

@inproceedings{verma2020correlated,
  title={Correlated Learning for Aggregation Systems},
  author={Verma, Tanvi and Varakantham, Pradeep},
  booktitle={Uncertainty in Artificial Intelligence},
  pages={60--70},
  year={2019},
  organization={PMLR}
}