Indian scientists are leveraging AI (artificial intelligence) to revolutionise material discovery. They employ techniques such as muti-property pre-training and transfer learning to achieve unparalleled accuracy in material properties forecasting. This breakthrough will significantly impact the development of advanced semiconductors and more efficient energy storage solutions.

Material science is fundamental for developing technologies ranging from semiconductor materials to renewable energy. However, traditional methods are often slow, labour-intensive and expensive. These experiential methods hinder the discovery of novel materials. Here, machine learning is a viable solution that allows accurate and rapid forecasting of material properties from existing data.

The IISC (Indian Institute of Science) researchers employed GNNs (Graph Neural Networks) to model intricate 3D structures of material. The team attained remarkable success in forecasting critical material properties by carefully optimising the GNN architecture and fine-tuning the size of training datasets.

IISc team, led by assistant professor Gopalakrishnan said, “AI enables us to leverage existing knowledge for discoveries, accelerating the pace of innovation and expanding the boundaries of possibility.”

The IISc research demonstrates how AI can drive collaborative efforts and address critical national needs. Leveraging AI, India is not only advancing material science but also solidifying its position as a leader in technology.

Source: https://indiaai.gov.in/article/ai-revolutionizing-material-discovery-indian-scientists-forecast-new-properties