On the Robustness of Deep Features for Audio Event Classification in Adverse Environments | Publicación