the slide is about Use of GNNs in Predicting the Expression of Superconductivity in Materials Through the Use of Crystal Unit Cells. It should clearly explain the fact that we tried multiple different models and finally landed on GIN as the best for the projects use case and explain why a GIN is best
the slide is about Use of GNNs in Predicting the Expression of Superconductivity in Materials Through the Use of Crystal Unit Cells. It should clearly explain the fact that we tried multiple different models and finally landed on GIN as the best for the projects use case and explain why a GIN is best
Created using ChatSlide
Using graph neural networks, we model crystal unit cells as graphs to predict superconductivity. After comparing MPNN, GCN, GAT, and GIN models, GIN achieved the highest accuracy due to its expressive power. Future work addresses dataset limitations and hyperparameter tuning, inviting collaboration on advanced GINs.