Phase Prediction of High-Entropy Alloys by Integrating Criterion and Machine Learning Recommendation Method

• DOI: 10.3390/ma15093321
• Publication Date: 2022-05-05T17:10:26Z
• Authors: Shuai Hou; Yujiao Li; Meijuan Bai; Mengyue Sun; Weiwei Liu; Chao Wang; Halil Tetik; Dong Lin
• Keywords: Quinary; High-Entropy Alloys; Interpretability

The comprehensive properties of high-entropy alloys (HEAs) are highly-dependent on their phases. Although a large number machine learning (ML) algorithms has been successfully applied to the phase prediction HEAs, accuracies among different ML based same dataset vary significantly. Therefore, selection an efficient algorithm would significantly reduce and cost experiments. In this work, HEAs (PPH) is proposed by integrating criterion recommendation method (MLRM). First, meta-knowledge table characteristics performance candidate established, meta-learning adopted recommend with desirable accuracy. Secondly, MLRM improved engineered more for prediction. Finally, considering poor interpretability generalization single algorithms, PPH combining advantages improve accuracy validated 902 samples from 12 datasets, including 405 quinary 359 senary 138 septenary HEAs. experimental results shows that achieves than traditional method. average in all, quinary, senary, 91.6%, 94.3%, 93.1%, 95.8%, respectively.