Machine learning for melt-pool prediction in metal additive manufacturing: source-aware evaluation, data quality, and comparison with analytical physics — reproducibility package

Résumé fourni par la source

Frozen reproducibility release corresponding to the manuscript: "Machine learning for melt-pool prediction in metal additive manufacturing: source-aware evaluation, data quality, and comparison with analytical physics." This release contains the final reproducibility notebooks and derived outputs for: source-aware and alloy-held-out evaluation physical data-quality auditing corrected Eagar-Tsai analytical benchmarking LOSO classification analysis corrected process maps melt-pool width validation uncertainty and empirical conformal analysis final source-held-out SHAP attribution analysis Eagar-Tsai-derived defect-mode sensitivity analysis The analysis uses the public MeltpoolNet dataset pinned to commit: 6d68e2acaad8d074134a7f3d843278649774d661

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.