Zero-shot Audio Classification Explanation on ESC50 White Noise contamination
32.97AI ScoreLMAC-ZS
Evaluation results
| Method | Links | ||||||||
|---|---|---|---|---|---|---|---|---|---|
| LMAC-ZSTraining Set=Full, Masking Domain=STFT-Masking, Classification Accuracy=57.4%2024.05 | 32.97 | 11.86 | 10.63 | 64 | 0.7 | 0.65 | 11.85 | 9 | |
| LMAC-ZSTraining Set=CT, Masking Domain=Mel-Masking, Classification Accuracy=65.2%2024.05 | 28.52 | 17.72 | 12.78 | 42 | 0.64 | 0.82 | 9.18 | 19 | |
| LMAC-ZSTraining Set=CT, Masking Domain=STFT-Masking, Classification Accuracy=57.4%2024.05 | 19.85 | 21.51 | 7.13 | 63 | 0.53 | 0.52 | 12.24 | 8 | |
| GradCam++Masking Domain=STFT-Masking, Classification Accuracy=57.4%2024.05 | 19.5 | 24.01 | 8.04 | 66 | 0.5 | 0.42 | 12.42 | 33 | |
| GradCamMasking Domain=STFT-Masking, Classification Accuracy=57.4%2024.05 | 14.92 | 31.89 | 5.95 | 66 | 0.32 | 0.77 | 11.4 | 12 | |
| LMAC-ZSTraining Set=Full, Masking Domain=Mel-Masking, Classification Accuracy=65.2%2024.05 | 14.25 | 27.92 | 6.62 | 41 | 0.42 | 0.86 | 8.86 | 11 | |
| IGMasking Domain=STFT-Masking, Classification Accuracy=57.4%2024.05 | 10.17 | 34.35 | 4.89 | 66 | 0.3 | 0.69 | 11.8 | 1.1 | |
| GradCam++Masking Domain=Mel-Masking, Classification Accuracy=65.2%2024.05 | 7.12 | 37.03 | 2.97 | 52 | 0.26 | 0.43 | 10.33 | 33.5 | |
| SmoothGradMasking Domain=STFT-Masking, Classification Accuracy=57.4%2024.05 | 7.1 | 36.53 | 2.66 | 66 | 0.25 | 0.52 | 12.15 | 0.04 | |
| GradCamMasking Domain=Mel-Masking, Classification Accuracy=65.2%2024.05 | 3.65 | 43.79 | 1.43 | 34 | 0.12 | 0.75 | 9.41 | 14 | |
| SmoothGradMasking Domain=Mel-Masking, Classification Accuracy=65.2%2024.05 | 1.72 | 47.93 | 0.56 | 4 | 0.04 | 0.28 | 10.54 | 3.5 | |
| IGMasking Domain=Mel-Masking, Classification Accuracy=65.2%2024.05 | 1.57 | 47.97 | 0.55 | 8.4 | 0.039 | 0.54 | 10.16 | 3.4 |