This is the repository for the collection of Graph-based Deep Learning for Communication Networks.

Need i say more? GNN-DRL routing inspired by https://ieeexplore.ieee.org/document/9680855 and built from work done here https://github.com/jwwthu/GNN-Communication-Networks

[PACT'24] GraNNDis. A fast and unified distributed graph neural network (GNN) training framework for both full-batch (full-graph) and mini-batch training. Provides unification of full-/mini-batch training using a novel data/communication structure.

Architecturally Aware Optimisation for Custom GNN-based Models in Communication Networks

Graph Neural Networks (GNNs) are a cutting-edge branch of machine learning that targets the processing of data structured in the form of graphs. This technique is particularly potent for applications where data is inherently relational and interconnected, such as social networks, chemical structures, or even communication networks.

This project aims to bridge the communication gap for the hearing and speech-impaired community by developing an intelligent system that recognizes hand gestures across multiple cultural sign languages. Tech Stack: Python | OpenCV | MediaPipe | PyTorch | TensorFlow | Graph Neural Networks (GNNs)

NexFlow AI: An Urban Nervous System for traffic. Using Graph Neural Networks (GNN) and Graph Attention (GAT), NexFlow solves hyper-dense Indian roadway congestion. It enables real-time junction communication and absolute emergency vehicle priority. Built with PyTorch Geometric and SUMO, it’s a production-ready, OpenEnv-compliant RL controller.