We want to investigate how pooling layers affect the performance of a ML pipeline. We want to investigate this within the citation network domain, more specifically we will use the three datasets: Cora, Pubmed and Citeseer. But we will start with one of them and hopefully expand our work to all three. The downstream task is node classification because all three datasets assigns a class to all nodes and a node classification algorithm could potentially help others in the future. All this means that we want to change the pooling layer in the gnn and see how this affects the node classification performance and concluded on that.
Initial Idea
We want to investigate how pooling layers affect the performance of a ML pipeline. We want to investigate this within the citation network domain, more specifically we will use the three datasets: Cora, Pubmed and Citeseer. But we will start with one of them and hopefully expand our work to all three. The downstream task is node classification because all three datasets assigns a class to all nodes and a node classification algorithm could potentially help others in the future. All this means that we want to change the pooling layer in the gnn and see how this affects the node classification performance and concluded on that.