Data-Guided Feature Identification for Predicting Specific Heat of Multicomponent Alloys

Understanding oxidation of Fe-Cr-Al alloys through explainable artificial intelligence

“…The prediction accuracies of the NN model are 100% for training set and 88.2% (15/17) for test set. The confusion matrix of the test set is [[4, 1, 0], [0, 9, 0], [0, 1,2]]. The F1 scores for spallation, good, and non-projective oxidation are 88.9%, 90%, and 80%.…”

“…Although no computational approach can displace the ground truth of experiments, the time and cost associated with only experimental studies to guide us toward best alloy composition can retard significant progress. With the exponential increase in demand for new materials, data-driven modeling based on experimental dataset coupled with expert domain knowledge has flourished in the last decade, [2] primarily from academic research. Although data-driven modeling (based on compositional features) has been applied to predict a variety of mechanical properties, [3][4][5][6] corrosion and oxidation behavior are often considered a challenging problem due to the inherent complexity of these physical processes.…”

Elucidating Precipitation in FeCrAl Alloys through Explainable AI: A Case Study

“…Among the three primary variables of the alloy composition, concentration of Cr has the highest importance on overall prediction of hardness change while the Al content demonstrated the lowest contribution among the alloying elements. Although, this standard visual of getting material insight from data is ubiquitous in many datadriven modeling publications [14,[60][61], little directional conclusion can be made from the feature importance plot. For example, the model tells us that Cr concentration is the most important feature among the alloying elements without giving us any information about whether it increases or decreases hardness change which is of importance to material scientists.…”

“…With the advancement and easy access to computing power, the widespread implementation of data-driven approaches has opened up a new and unexplored tool for understanding α'-precipitation in FeCrAl. Such tools have been ubiquitously used in material science, from property prediction to process parameter optimization [14][15][16]. For predicting hardness in alloy based on composition and condition, researchers have been progressively using advanced algorithms such as neural network (NN), gradient boost regression (GBR), Support Vector Machine (SVM) [17] to predict the hardness of alloys.…”