Skip to content

Semi-Supervised End-to-End Learning for Integrated Sensing and Communications

License

Notifications You must be signed in to change notification settings

josemateosramos/SSLISAC

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

22 Commits
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Semi-Supervised End-to-End Learning for Integrated Sensing and Communications

Getting Started

This code is based on Pytorch 1.12.1 and CUDA 11.3.1, and may not work with other versions. For more information about how to install these versions, check the Pytorch documentation.

The simulation parameters to train and test different scenarios are located in the simulation_parameters.py file within the lib/ directory. The methods/ directory contains the scripts to train and test all methods: (i) baseline (ii) supervised learning, (iii) unsupervised learning, and (iv) semi-supervised learning. To obtain Fig. 4 of the original paper, semi-supervised learning should be run with different supervised training iterations.

Additional information

If you decide to use the source code for your research, please make sure to cite our paper:

  • J. M. Mateos-Ramos, B. Chatelier, C. Häger, M. F. Keskin, L. L. Magoarou, and H. Wymeersch, "Semi-Supervised End-to-End Learning for Integrated Sensing and Communications," in IEEE International Conference on Machine Learning for Communication and Networking (ICMLCN), Stockholm, Sweden, 2024.