Prediction of infinite dilution activity coefficients in polymer solutions using Graph Neural Networks.
The project aims to predict infinite dilution activity coefficients in polymer solutions by utilizing Graph Neural Networks, providing information on comparing the predictions of Mechanistic and other data-driven models for the infinite dilution activity coefficients. The prediction of intended attributes is carried out through the use of Graph Neural Networks, Feed Forward Neural Networks, and Random Forest models, and their performance is compared. The results indicate that the GNN model achieves better performance than the other models. Additionally, a comparison is made between the performance of the GNN model and mechanistic models to validate its performance.