Machine learning-accelerated discovery and optimization of thermoelectric materials for energy conversion

Recently, machine learning (ML) has received growing attention as a powerful tool for accelerating the discovery and optimization of thermoelectric materials. By utilizing large-scale datasets generated from high-throughput calculations and experimental measurements, ML models are able to efficiently predict key properties, identify promising candidates, and guide material design. Herein we summarize the latest progress in this rapidly evolving field and present a comprehensive overview on typical ML workflows, key algorithms and efficient models for predicting thermoelectric properties, including both indirect and direct predictions of electrical and thermal transport characteristics. Subsequently, we discuss how ML facilitates the discovery of novel thermoelectric materials and supports the optimization of composition, synthesis processes, and microstructures. Finally, we explore current challenges and future opportunities for integrating ML into thermoelectric investigation. This review aims to provide a comprehensive understanding of ML applications in thermoelectrics, thereby highlighting their potential to accelerate thermoelectric materials development for energy conversion and waste-heat recovery applications.