Medley-Solos-DB results
Baselines
Our baselines consist on random samples from the training set of the Medley-solos-DB dataset and prototypes obtained with the APNet model.
| class | dataset sample | APNet |
|---|---|---|
| clarinet | ||
| distorted electric guitar | ||
| female singer | ||
| flute | ||
| piano | ||
| tenor saxophone | ||
| trumpet | ||
| violin |
Our models
We show the results obtained with PECMAE-3 (3 prototypes per target class). For each class, we sonify two of the prototypes.
| class | PECMAE-3 (prototype 0) | PECMAE-3 (prototype 1) |
|---|---|---|
| clarinet | ||
| distorted electric guitar | ||
| female singer | ||
| flute | ||
| piano | ||
| tenor saxophone | ||
| trumpet | ||
| violin |
Prototype-class connections
In PECMAE, prototypes are linearly connected to the classification layer.
The following plot shows the weights learned for these connections.
Note that certain prototypes may have a positive correlation with related classes.
For example, flute and clarinet.
