# Secondary labels are effective for bronze medal (76th solution)

Thank you for the interesting competition, and congratulations to the winning teams!
This is our 76th (Private: 0.6109, Public: 0.6861) solution.
Our solution summary
(1) Baseline framework is kkiller’s public notebook (Thanks to
[@kneroma](https://www.kaggle.com/kneroma)
!)
[https://www.kaggle.com/kneroma/clean-fast-simple-bird-identifier-training-colab](https://www.kaggle.com/kneroma/clean-fast-simple-bird-identifier-training-colab)
(2) Just adding secondary labels information
primary_label: 0.995
secondary_labels: 0.5 (Using 0.7 or 0.995 is also same performance)
others: 0.0025
(3) Choose threshold that is optimal for train soundscapes
We also tried data augmentation (noise, mixup … ), test time augmentation, and so on. However, these things did not work in our environment.
Thank you!