Accelerated discovery of oxidation-resistant ultra-high temperature ceramics via data driven methodology

To accelerate the discovery of ultra-high temperature ceramics (UHTCs) with excellent oxidation resistance at high temperatures, this paper is the first attempt to apply a high-throughput experiments (HTEs) assisted data -driven strategy in the prediction of oxidation recession of transition metal diboride-SiC ceramic composites. The data generated by HTEs enables us to apply the machine learning (ML) method in discovering novel oxidationresistant UHTCs. Artificial Neural Network (ANN) and Kernel logistic regression (KLR) models show an outstanding performance against other ML models. Optimum compositions with exceptional oxidation resistance were rapidly observed using the trained KLR model.