Skip to content

This repository implements recurrent and graph neural networks from scratch using PyTorch. It covers LSTM and GRU models for image classification on FashionMNIST, and a 2-layer GCN for node classification on the Cora citation network. It also explores the impact of heterophily on GCN performance.

Notifications You must be signed in to change notification settings

A1SHWARYANAYAK/Recurrent-and-Graph-Neural-Networks

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 

Repository files navigation

Recurrent-and-Graph-Neural-Networks

This project provides a hands-on implementation of Recurrent Neural Networks (RNNs) and Graph Convolutional Networks (GCNs) using PyTorch. It covers:

  • LSTM and GRU for Classification: Implements LSTM and GRU models for image classification on the FashionMNIST dataset.
  • GCN for Node Classification: Implements a 2-layer GCN for node classification on the Cora citation network dataset. Explores the concept of heterophily and its impact on GCN performance.
  • Graph Autoencoder (GAE) for Link Prediction: Implements a GAE using a GCN encoder for link prediction on the Cora dataset.

This project is ideal for anyone learning about RNNs, GCNs, and their applications in various machine learning tasks.

Usage

  1. LSTM/GRU for Classification:
    • Run the code in section 1 to train and evaluate LSTM and GRU models on the FashionMNIST dataset.
  2. GCN for Node Classification:
    • Run the code in section 2 to train and evaluate a GCN model on the Cora dataset.
    • Explore the impact of heterophily by running the code in Task 10.
  3. GAE for Link Prediction:
    • Run the code in Task 11 to train and evaluate a GAE for link prediction on the Cora dataset.

Key Features

  • RNN Implementations: Provides implementations of LSTM and GRU models from scratch.
  • GCN Implementation: Implements a 2-layer GCN from scratch.
  • Heterophily Analysis: Explores the concept of heterophily and its impact on GCN performance.
  • Link Prediction with GAE: Demonstrates how to use a GCN encoder for link prediction in a GAE.
  • Clear Explanations: Includes comments and explanations within the code to guide understanding.
  • Visualization: Uses matplotlib to visualize training progress and results.

Contributor

AISHWARYA NAYAK (Contributions are welcome! Feel free to open issues or submit pull requests.)

About

This repository implements recurrent and graph neural networks from scratch using PyTorch. It covers LSTM and GRU models for image classification on FashionMNIST, and a 2-layer GCN for node classification on the Cora citation network. It also explores the impact of heterophily on GCN performance.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published