This review discusses the application of machine learning (ML) in the design of high-entropy alloys (HEAs), focusing on common and advanced ML models and algorithms used in HEA design. HEAs are alloys with four or more elements, each with a high atomic percentage, and they exhibit excellent mechanical, thermal, and corrosion resistance properties. However, traditional experimental methods for designing HEAs are inefficient and costly. High-throughput calculations (HTC) have improved the efficiency of HEA design, but their accuracy is limited due to indirect correlations between theoretical calculations and performance. ML, which uses real data, has gained attention for its ability to predict material properties and assist in design. The review analyzes the advantages and limitations of various ML models and algorithms, such as neural networks (NNs), support vector machines (SVMs), Gaussian processes (GPs), k-nearest neighbors (KNNs), and random forests (RFs), and discusses their potential weaknesses and optimization strategies. It also explores advanced ML models, including active learning (AL), genetic algorithms (GAs), deep learning (DL), and transfer learning (TL), which offer improved performance in HEA design. The review emphasizes the importance of data quality, model interpretability, and the integration of computational theory with experimental validation for effective HEA design. The key challenges include data dependence, model complexity, generalization, and interpretability, which require optimization strategies to enhance the accuracy and reliability of ML models in HEA design.This review discusses the application of machine learning (ML) in the design of high-entropy alloys (HEAs), focusing on common and advanced ML models and algorithms used in HEA design. HEAs are alloys with four or more elements, each with a high atomic percentage, and they exhibit excellent mechanical, thermal, and corrosion resistance properties. However, traditional experimental methods for designing HEAs are inefficient and costly. High-throughput calculations (HTC) have improved the efficiency of HEA design, but their accuracy is limited due to indirect correlations between theoretical calculations and performance. ML, which uses real data, has gained attention for its ability to predict material properties and assist in design. The review analyzes the advantages and limitations of various ML models and algorithms, such as neural networks (NNs), support vector machines (SVMs), Gaussian processes (GPs), k-nearest neighbors (KNNs), and random forests (RFs), and discusses their potential weaknesses and optimization strategies. It also explores advanced ML models, including active learning (AL), genetic algorithms (GAs), deep learning (DL), and transfer learning (TL), which offer improved performance in HEA design. The review emphasizes the importance of data quality, model interpretability, and the integration of computational theory with experimental validation for effective HEA design. The key challenges include data dependence, model complexity, generalization, and interpretability, which require optimization strategies to enhance the accuracy and reliability of ML models in HEA design.