Different applications of machine learning approaches in materials science and engineering: Comprehensive review

Over the last decades, considerable advancements in artificial intelligence (AI) approaches have eventuated in their extensive applications in all scientific scopes such as materials science and engineering (MSE). One of the most promising AI-based techniques that have recently revolutionized the MSE toolbox is machine learning (ML). The industrial/commercial application of ML techniques has demonstrated great potential to accelerate both fundamental and applied studies via learning rules from datasets and developing predictive models. The prominent objective of this review investigation is to discuss the recent development of machine learning (ML) principles, algorithms and databases in materials science and engineering (MSE). As the novelty, the role of various prevalent ML-based techniques (i.e., Support vector machine (SVM), Naive Bayes classifier (NBC), Decision tree (DT), Artificial neural network (ANN) and Deep learning (DL)) for the characterization and discovery of stable materials is aimed to be reviewed. The authors have made their efforts to investigate the replacement feasibility of first-principle techniques with ML. Additionally, the efficiency of different ML-based algorithms in different scopes of MSE like the biomaterials' characterization, nanomaterials analysis, inverse design, drug discovery and estimation of the physicochemical properties are discussed, comprehensively. Ultimately, possible solutions and future research paths toward solving the challenges in computational MSE are proposed.