A Study on Transfer Learning for Acoustic Event Detection in a Real Life Scenario

P. Arora, R. Haeb-Umbach, in: IEEE 19th International Workshop on Multimedia Signal Processing (MMSP), 2017.

Conference Paper | English
Author
Abstract
In this work, we address the limited availability of large annotated databases for real-life audio event detection by utilizing the concept of transfer learning. This technique aims to transfer knowledge from a source domain to a target domain, even if source and target have different feature distributions and label sets. We hypothesize that all acoustic events share the same inventory of basic acoustic building blocks and differ only in the temporal order of these acoustic units. We then construct a deep neural network with convolutional layers for extracting the acoustic units and a recurrent layer for capturing the temporal order. Under the above hypothesis, transfer learning from a source to a target domain with a different acoustic event inventory is realized by transferring the convolutional layers from the source to the target domain. The recurrent layer is, however, learnt directly from the target domain. Experiments on the transfer from a synthetic source database to the reallife target database of DCASE 2016 demonstrate that transfer learning leads to improved detection performance on average. However, the successful transfer to detect events which are very different from what was seen in the source domain, could not be verified.
Publishing Year
Proceedings Title
IEEE 19th International Workshop on Multimedia Signal Processing (MMSP)
LibreCat-ID

Cite this

Arora P, Haeb-Umbach R. A Study on Transfer Learning for Acoustic Event Detection in a Real Life Scenario. In: IEEE 19th International Workshop on Multimedia Signal Processing (MMSP). ; 2017.
Arora, P., & Haeb-Umbach, R. (2017). A Study on Transfer Learning for Acoustic Event Detection in a Real Life Scenario. In IEEE 19th International Workshop on Multimedia Signal Processing (MMSP).
@inproceedings{Arora_Haeb-Umbach_2017, title={A Study on Transfer Learning for Acoustic Event Detection in a Real Life Scenario}, booktitle={IEEE 19th International Workshop on Multimedia Signal Processing (MMSP)}, author={Arora, Prerna and Haeb-Umbach, Reinhold}, year={2017} }
Arora, Prerna, and Reinhold Haeb-Umbach. “A Study on Transfer Learning for Acoustic Event Detection in a Real Life Scenario.” In IEEE 19th International Workshop on Multimedia Signal Processing (MMSP), 2017.
P. Arora and R. Haeb-Umbach, “A Study on Transfer Learning for Acoustic Event Detection in a Real Life Scenario,” in IEEE 19th International Workshop on Multimedia Signal Processing (MMSP), 2017.
Arora, Prerna, and Reinhold Haeb-Umbach. “A Study on Transfer Learning for Acoustic Event Detection in a Real Life Scenario.” IEEE 19th International Workshop on Multimedia Signal Processing (MMSP), 2017.

Link(s) to Main File(s)
Access Level
Restricted Closed Access
External material:
Supplementary Material
Description
Poster

Export

Marked Publications

Open Data LibreCat

Search this title in

Google Scholar