The goal of this project is to provide a theoretic analysis, explaining the effect of homophily on Graph Neural Network Performace, that leverages on recent results in random family networks and their geodesics, by exploring the structural effects of homophily.
The goal of this project is to provide a theoretic analysis, explaining the effect of homophily on Graph Neural Network Performace, that leverages on recent results in random family networks and their geodesics, by exploring the structural effects of homophily. The theory developed can hopefully be used to build better rewiring techniques and other methods to alleviate harmful heterophily.
The goal of this project is to provide a theoretic analysis, explaining the effect of homophily on Graph Neural Network Performace, that leverages on recent results in random family networks and their geodesics, by exploring the structural effects of homophily.