Answers are grounded in this paper and cite their sources.
Computational study demonstrates accurate hardness extrapolation in multi-component alloys using symbolic regression, highlighting an effective method to overcome small-data constraints.
Data‐driven materials science integrates domain knowledge with machine learning (ML), yet ML's extrapolation limitations on small datasets are often overlooked. This study develops a comprehensive Symbolic Regression Extrapolation Framework (SREF) for the integrated property and component design of alloys. Evaluating 6 ML algorithms on an optimized dataset of 273 hardness values, the symbolic regression model demonstrates exceptional out‐of‐domain generalizability. Incorporating high‐hardness data points, or virtual samples generated via symbolic regression, further enhances accuracy. An ensemble of four symbolic regression models designed a high‐entropy alloy (HEA), Al 49 Co 14 Cr 29 Fe 5 Ni 1 Cu 2 , achieving a hardness of 863.5 HV, exceeding the original dataset's upper limit. Furthermore, by modeling composition‐derived alloy meta‐features, a novel seven‐component HEA, Al 40 Co 12 Cr 22 Fe 13 Ni 2 Cu 6 (Mo/Ti/V) 5 , was developed. All synthesized alloys exhibit hardness values over 750 HV, with prediction errors below 1.0%. This methodology effectively overcomes small‐data constraints for discovering high‐performance multi‐component alloys.
No takes yet. Share an insight, caveat, or question.
Yu et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: