Link-Based Classification Using Graph Neural Networks

Graph neural networks (GNNs) are a class of deep learning models that capture intrinsic data patterns to facilitate model training. They are finding extensive applications in social and ego networks, molecular discovery, and other domains where the data has an underlying graph structure.

In this project, we will develop a graph convolutional network (GCN) to classify the scientific publications in the Cora dataset. As the Cora dataset consists of interlinked data, using GNNs will allow us to capture more data correlations as compared to conventional neural networks for improved model performance. We will import the Cora dataset, implement the graph convolutional network, and use it to classify the scientific publications in the Cora dataset. Moreover, we will analyze the model performance for different split ratios of the dataset.