Deep Learning Models for Structural-Property Data

Deep Learning models are inspired by the neural networks found in the human brain. They excel in handling extensive datasets and deciphering intricate patterns. Selecting the ideal model that achieves the optimal balance between accuracy and efficiency is crucial when utilizing data-driven methods for property prediction. Ongoing research in this direction:

  • Artificial Neural Networks (ANNs) for Interatomic Potential (MLIP) training for Carbon-based Materials.
  • Graph Neural Networks (GNNs) for materials and molecular properties predictions.