torchgeometric

Make GNN easy to start with

A Pytorch based implementation of classical GNNs.

This project aims to predict the satisfiability of a SAT3 problem instance using GNNs or LSTMs. Since the SAT3 problem is similar to the 3-coloring, it attempts to solve that as well.

Implementation of a Graph Echo State Network (GraphESN) using PyTorch Geometric (PyG).

Semi-Supervised Mean Teacher framework with Attentive GINE for molecular property prediction (SIDER, Tox21, PCBA). 02456 Deep Learning DTU Project.

DistrGNN is a project supported by a scholarship from the University of Pisa, focused on exploring and implementing model parallelism for Graph Neural Networks (GNNs). The goal is to investigate methods for efficiently distributing GNN computations across multiple computing nodes, optimizing performance and scalability.

This project implements models for Protein-Protein Interaction (PPI) prediction, focusing on graph-based methods such as the SOTA GNN model call GIPA.

Knowledge graph and GNN benchmark for programmable genome-writing enzymes, with ESM-2 embeddings, GraphSAGE/GAT link prediction, and an editor-selection API.

PyTorch Implementation of "Metropolis-Hastings Data Augmentation for Graph Neural Networks". NeurIPS 2021.

Uncertainty-aware deep learning ensemble that turns raw Sentinel-2 imagery into georeferenced land-cover maps with per-pixel confidence - five fused segmentation models, 27 benchmarked experiments, and a decade of measured urban expansion over Vilnius.

This repository presents and compares HeterSUMGraph and variants doing extractive summarization, named entity recognition or both. HeterSUMGraph and variants use GATv2Conv (from torch_geometric).