AI for Physics & Chemistry

• Design experiments to benchmark and assess SOTA and commonly used MLFF (machine learning force field) architectures, including MACE, QuinNet.
• Explore the importances of LR in different molecular systems using both ML driven and physics driven analyses.

AI for Biology

• Leveraging multi-task learning for protein function prediction via heterogeneous graphs
• Integration of diverse protein datasets to enhance multi-task learning capabilities, enriching the model with additional research data
• Conducting single-task experiments and comparing their results with those of the multi-task learning approach

Medical Image segmentation

• Segmentation of bone rings in CT images using traditional image segmentation methods and neural network methods