This method is not used in the our system in latest version. There is a way to perform better with smaller models. We also use this method. We have applied MTST to the above method, however, we found that it causes performance degradation. Therefore, it is no longer supported. As a reference, this method has been developed in January 2019. It is technically insufficient because the paper was published after long time spend for user experiments rather than method of a similar published date. Therefore, it is recommended to use the latest method rather than our methods.
* MTST
- ESC-50-master(Dataset)
+ weights
+ fold1
- weight.pth
+ result.out
+ fold2
...
+ fold5
+ test.py
Fold | Accuracy | Recall | Precision | Weighted F1 | MCC | AUC |
---|---|---|---|---|---|---|
1 | 81.2% | 81.2% | 84.4% | 81.1% | 81.0% | 90.4% |
2 | 81.0% | 81.0% | 83.5% | 80.4% | 80.7% | 90.3% |
3 | 81.2% | 81.2% | 82.6% | 80.7% | 80.9% | 90.4% |
4 | 84.8% | 84.8% | 85.7% | 84.2% | 83.7% | 91.8% |
5 | 81.2% | 81.2% | 84.0% | 81.1% | 80.9% | 90.4% |
avg. | 81.9% | 81.9% | 84.0% | 81.5% | 81.4% | 90.7% |