Awesome

Globecom2020-ResourceAllocationGNN

Code for Globecom2020 paper: Resource Allocation based on Graph Neural Networks in Vehicular Communications

The paper is available online: https://ieeexplore.ieee.org/abstract/document/9322537

Code is based on https://github.com/CooperLWang/Learn-CompressCSI-RA-V2X-Code

Learning Environment:

(1) Keras 2.2.4 (2) TensorFlow 1.14.0

Why obtain different figures from figures in the paper when running the code?

The code for figures only plots for original results data without smoothing steps so that you may see a different figure from the clean figures in the paper (which smooths the return over adjacent episodes for clarity in demonstration).

You can smooth the returns by yourself with the saved results.