Table 8 Explainability performance comparison on the Fitzpatrick17k-C dataset.
From: MTAKD: multi-teacher agreement knowledge distillation for edge AI skin disease diagnosis
Framework | Insertion AUC | Deletion AUC | Insertion \(\:\varvec{p}\)-value vs. MTAKD | Deletion \(\:\varvec{p}\)-value vs. MTAKD |
|---|---|---|---|---|
Hinton-KD 8 | 0.0794 ± 0.0081 | 0.1049 ± 0.0126 | 0.0002 | 0.0016 |
AT-KD 20 | 0.0812 ± 0.0042 | 0.1092 ± 0.0035 | 0.0001 | 0.0001 |
RKD 16 | 0.0798 ± 0.0158 | 0.1054 ± 0.0163 | 0.0008 | 0.0024 |
SSD-KD 10 | 0.0792 ± 0.0109 | 0.1081 ± 0.0102 | 0.0003 | 0.0010 |
Averaging-KD 12 | 0.1339 ± 0.0055 | 0.0804 ± 0.0061 | 0.0001 | 0.0007 |
CA-MKD 13 | 0.1110 ± 0.0029 | 0.1046 ± 0.0026 | 0.0001 | 0.0001 |
MTAKD | 0.1724 ± 0.0080 | 0.0582 ± 0.0031 | - | - |
