Researchers at Indian Institute of Science (IISc), in collaboration with UCL University College London, have developed a revolutionary machine learning (ML) approach that predicts material properties using limited data, paving the way for advancements in semiconductors and energy storage.
Led by Sai Gautam Gopalakrishnan, Assistant Professor at IISc’s Department of Materials Engineering, the team employed transfer learning and optimized Graph Neural Networks (GNNs) to make accurate predictions about material properties like the piezoelectric coefficient and band gap values, even for materials the model hadn’t seen before.
Their Multi-property Pre-Training (MPT) framework is a game-changer, boosting predictive accuracy by training the model on seven bulk material properties simultaneously. This innovation has far-reaching implications, from transforming India’s semiconductor manufacturing ambitions to enhancing battery technology by predicting ion mobility within electrodes.
Read the article to know more. Link in the comment 🔗
Led by Sai Gautam Gopalakrishnan, Assistant Professor at IISc’s Department of Materials Engineering, the team employed transfer learning and optimized Graph Neural Networks (GNNs) to make accurate predictions about material properties like the piezoelectric coefficient and band gap values, even for materials the model hadn’t seen before.
Their Multi-property Pre-Training (MPT) framework is a game-changer, boosting predictive accuracy by training the model on seven bulk material properties simultaneously. This innovation has far-reaching implications, from transforming India’s semiconductor manufacturing ambitions to enhancing battery technology by predicting ion mobility within electrodes.
Read the article to know more. Link in the comment 🔗
