Phase diagram determination using machine learning methods in materials science and engineering: a review

Machine learning methods are widely applied in various areas of materials science and engineering. They prove to be a powerful framework for understanding structure–property relationships and aid in discovering novel materials. In this review, we address various machine learning approaches used to determine the phase diagram of different material systems. We provide a detailed discussion on data generation methods, preprocessing, and common data operations that are essential for model development. We examine unsupervised and supervised machine learning methods, along with their training, testing, and validation processes. Furthermore, we highlight the challenges in phase diagram calculation and propose potential solutions. Finally, we discuss the applications of different machine learning methods in constructing phase and transformation diagrams.