khames-lab / node-embedding-gcn-karate

📌Graph Convolutional Network (GCN) on Zachary's karate club network

Graph Convolutional Network (GCN) for Karate Club Dataset

This repository contains code for training a Graph Convolutional Network (GCN) on Zachary's karate club network and visualizing node embeddings in a 2D space.

Introduction

The Karate Club dataset is a classic social network dataset representing the interactions between members of a karate club. This project uses PyTorch Geometric to build a GCN model and visualize the embeddings of the karate club members.

Requirements

• Python 3.x
• PyTorch
• PyTorch Geometric
• NetworkX
• Matplotlib
• Scikit-Learn

Install the required dependencies using the following command:

Model Architecture

The GCN model consists of two graph convolutional layers with hyperbolic tangent (tanh) activation functions and a linear classifier. The model is trained using Cross-Entropy Loss and the Adam optimizer.

The script will train the model and generate a scatter plot of node embeddings using t-SNE. The plot will be displayed or saved as an image file.