Table 2 Comparison of models’ prediction accuracy on unseen alloy systems.
From: A general and transferable deep learning framework for predicting phase formation in materials
Alloy system | SNN3 | SNN4 | CNN1 | CNN2 | CNN3 |
|---|---|---|---|---|---|
Leave-one system-out test on 160 ternary systems | – | 76.5% | – | 77.3% | 83.8% |
Leave-one system-out test on Al–Ni–Zr system | 62.3% | 63.5% | 68.4% | 66.4% | 80.3% |
Ir–Ni–Ta–(B), Mg–Cu–Yb, S-bearing BMGs | 17/28 | 9/28 | 18/28 | 23/28 | 27/28 |
RE6Fe72B22 | 5/13 | 8/13 | 7/13 | 8/13 | 10/13 |
Binary alloys outliers | 8/18 | 16/18 | 13/18 | 13/18 | 16/18 |
