Table 4 PSDS hyperparameters
From: Multi-encoder attention-based architectures for sound recognition with partial visual assistance
Hyperparameter | PSDS1 | PSDS2 |
|---|---|---|
Detection tolerance criterion | 0.7 | 0.1 |
Ground truth intersection criterion | 0.7 | 0.1 |
Cross-trigger tolerance criterion | N/A | 0.3 |
Cost of class instability | 1 | 1 |
Cost of cross-triggers | 0 | 0.5 |
Maximum false positive rate | 100 | 100 |