This review article by Sijia Liu and Chao Yang from Shanghai Jiao Tong University, China, focuses on the application of machine learning (ML) in the design of high-entropy alloys (HEAs). HEAs, known for their excellent properties and wide compositional space, are challenging to design through traditional experimental methods due to low efficiency and high costs. High-throughput calculation (HTC) has improved design efficiency but remains limited by the indirect correlation between theoretical calculations and performance. ML, using real data, has emerged as a promising tool to assist in material design, offering advantages in capturing complex patterns and non-linear relationships. The article reviews common and advanced ML models and algorithms used in HEA design, including neural networks (NNs), support vector machines (SVMs), Gaussian processes (GPs), k-nearest neighbors (KNN), and random forests (RFs). Each model's strengths and limitations are discussed, highlighting the need for effective data acquisition, utilization, and generation. Advanced ML models such as active learning (AL), genetic algorithms (GAs), deep learning (DL), and transfer learning (TL) are also explored, emphasizing their adaptability to small datasets and their potential in optimizing HEA design. The review identifies key challenges in ML models, including data dependence, model complexity, generalization, and interpretability. Strategies to address these issues, such as data preprocessing, model simplification, and interpretability techniques, are presented. The integration of computational theory and experimental validation is emphasized to enhance model accuracy and reliability. Finally, the article concludes by discussing the future prospects of ML in HEA design, highlighting the need for further research to fully leverage advanced tools and methods in HEA discovery and mechanistic exploration.This review article by Sijia Liu and Chao Yang from Shanghai Jiao Tong University, China, focuses on the application of machine learning (ML) in the design of high-entropy alloys (HEAs). HEAs, known for their excellent properties and wide compositional space, are challenging to design through traditional experimental methods due to low efficiency and high costs. High-throughput calculation (HTC) has improved design efficiency but remains limited by the indirect correlation between theoretical calculations and performance. ML, using real data, has emerged as a promising tool to assist in material design, offering advantages in capturing complex patterns and non-linear relationships. The article reviews common and advanced ML models and algorithms used in HEA design, including neural networks (NNs), support vector machines (SVMs), Gaussian processes (GPs), k-nearest neighbors (KNN), and random forests (RFs). Each model's strengths and limitations are discussed, highlighting the need for effective data acquisition, utilization, and generation. Advanced ML models such as active learning (AL), genetic algorithms (GAs), deep learning (DL), and transfer learning (TL) are also explored, emphasizing their adaptability to small datasets and their potential in optimizing HEA design. The review identifies key challenges in ML models, including data dependence, model complexity, generalization, and interpretability. Strategies to address these issues, such as data preprocessing, model simplification, and interpretability techniques, are presented. The integration of computational theory and experimental validation is emphasized to enhance model accuracy and reliability. Finally, the article concludes by discussing the future prospects of ML in HEA design, highlighting the need for further research to fully leverage advanced tools and methods in HEA discovery and mechanistic exploration.