Fig 1.
Node trend sequence mismatch problem between B(t) and H(t).
Fig 2.
Statistical analysis of MagNet dataset.
Fig 3.
MagNet data distribution illustration.
Fig 4.
PCA and t-SNE dimensionality reduction distribution of MagNet dataset samples.
Fig 5.
Comparison of material and temperature characteristics.
Fig 6.
Overall EMA-Mamba model diagram.
Table 1.
EMA-Mamba code flow.
Fig 7.
Memory augmentation mechanism model diagram.
Fig 8.
Illustration of B(t) and H(t) waveform inconsistency.
Table 2.
EMA-Mamba hyperparameter table.
Table 3.
Experimental environment.
Table 4.
K-fold experimental verification table.
Fig 9.
Prediction performance analysis of EMA-Mamba for N87 material at 25°C.
Table 5.
Leave-one-material-out cross-validation results (LOMO-CV).
Table 6.
Top-K ablation sensitivity analysis (N87 and 3C90 materials mixed temperature).
Table 7.
Full material experimental results.
Table 8.
LOTO-CV results on N87 and 3C90 materials.
Table 9.
Frequency range OOD test results.
Fig 10.
Comprehensive evaluation of EMA-Mamba temperature robustness.
Table 10.
Quantitative results of ablation experiments.
Table 11.
Comprehensive performance comparison of different neural network methods.
Table 12.
Statistical significance testing (Paired t-test based on 5-fold cross-validation).