Interpretation for Audio Classification on UrbanSound8K STFT-Masking, LJ-Speech contamination (test)
38.74Attribution Importance (AI)GradCam++
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| GradCam++mode=zero-shot2024.05 | 38.74 | 11.53 | 17.95 | 76 | 76 | 0.37 | 12.47 | 40 | |
| LMAC-ZSmode=zero-shot, variant=CT2024.05 | 35.96 | 15.91 | 18.33 | 68 | 67 | 0.63 | 11.92 | 7 | |
| IGmode=zero-shot2024.05 | 34.57 | 20.43 | 26.1 | 69 | 60 | 0.74 | 11.59 | 1 | |
| SmoothGradmode=zero-shot2024.05 | 34.35 | 19.43 | 24.32 | 76 | 62 | 0.52 | 12.11 | 0 | |
| LMAC-ZSmode=zero-shot, variant=Full2024.05 | 32.51 | 13.79 | 15.77 | 72 | 74 | 0.79 | 10.99 | 2 | |
| GradCammode=zero-shot2024.05 | 21.48 | 28.71 | 14.13 | 76 | 45 | 0.69 | 11.74 | 19 |