SE157: Accelerating materials discovery and synthesis through machine learning

Advanced materials underpin modern technology and sustainable development. Yet, accelerating the discovery of multi-element compounds remains a formidable challenge, constrained by the combinatorial explosion of the composition-structure-property paradigm. We developed a machine learning-guided framework that integrates advanced computational methods, databases, high-performance computing, and experimental validation. This AI/ML-driven approach dramatically speeds up the exploration of complex multi-element phase space and enhances the efficiency of structure searches and ab initio calculations, thus accelerating the discovery and synthesis of new materials. In this talk, we present our discovery of high-performance, rare-earth-free magnetic materials and novel complex quantum materials. We also examine the limitations of current machine learning approaches for materials discovery and discuss the challenges of bridging the gap between computational predictions and laboratory synthesis.