Machine learning in conductivity development of specific alloys

Although data-driven materials research is progressing rapidly, most contemporary review studies analyse machine learning applications from a broad materials perspective, offering limited quantitative assessment of the advancement of conductivity-oriented alloys. This research provides a comprehensive and conductivity-centric evaluation of machine learning and deep learning applications in metallic alloys, based on a systematic review of more than 200 published studies from 2019 to 2025. The originality of this study lies in its quantitative and functional classification of machine learning-driven research across the entire conductivity-focused alloy design pipeline, including conductivity prediction, accelerated alloy discovery, optimisation of alloy composition and processing, analysis of structure – property relationships, data-driven experimental design, and industrial applications. Statistical analysis reveals that artificial neural networks (22%), support vector machines (20%), and random forest models (18%) are the most prevalent in current conductivity prediction research, while innovative techniques such as Gaussian process regression, gradient boosting, and deep learning architectures are attracting increasing attention. This review emphasises critical challenges, including limited dataset sizes, insufficient representation of processing and microstructural characteristics, and the lack of uncertainty quantification. Based on these results, we suggest some new research directions that will help us use data to explore and improve high-conductivity metallic alloys.