english Icono del idioma   español Icono del idioma  

Please use this identifier to cite or link to this item: https://hdl.handle.net/20.500.12008/29961 How cite
Title: Toward interpretable polyphonic sound event detection with attention maps based on local prototypes
Authors: Zinemanas, Pablo
Rocamora, Martín
Fonseca, Eduardo
Font, Frederic
Serra, Xavier
Type: Ponencia
Keywords: Interpretability, Sound event detection, Prototypes
Issue Date: 2021
Abstract: Understanding the reasons behind the predictions of deep neural networks is a pressing concern as it can be critical in several application scenarios. In this work, we present a novel interpretable model for polyphonic sound event detection. It tackles one of the limitations of our previous work, i.e. the difficulty to deal with a multi-label setting properly. The proposed architecture incorporates a prototype layer and an attention mechanism. The network learns a set of local prototypes in the latent space representing a patch in the input representation. Besides, it learns attention maps for positioning the local prototypes and reconstructing the latent space. Then, the predictions are solely based on the attention maps. Thus, the explanations provided are the attention maps and the corresponding local prototypes. Moreover, one can reconstruct the prototypes to the audio domain for inspection. The obtained results in urban sound event detection are comparable to that of two opaque baselines but with fewer parameters while offering interpretability.
Publisher: Universitat Pompeu Fabra
IN: 6th Workshop on Detection and Classification of Acoustic Scenes and Events, DCASE 2021, Barcelona, Spain, 15-19 nov. 2021, pp. 50-54.
Citation: Zinemanas, P., Rocamora, M., Fonseca, E. y otros. Toward interpretable polyphonic sound event detection with attention maps based on local prototypes [en línea]. EN: 6th Workshop on Detection and Classification of Acoustic Scenes and Events, DCASE 2021, Barcelona, Spain, 15-19 nov. 2021, pp. 50-54.
License: Licencia Creative Commons Atribución - No Comercial - Sin Derivadas (CC - By-NC-ND 4.0)
Appears in Collections:Publicaciones académicas y científicas - Instituto de Ingeniería Eléctrica

Files in This Item:
File Description SizeFormat  
ZRFFS21.pdfVersión publicada723,73 kBAdobe PDFView/Open


This item is licensed under a Creative Commons License Creative Commons