Iterative Knowledge Distillation in R-CNNS for Weakly-Labeled Semi-Supervised Sound Event Detection

Khaled Koutini; Hamid Eghbal-Zadeh; Gerhard Widmer

Iterative Knowledge Distillation in R-CNNS for Weakly-Labeled Semi-Supervised Sound Event Detection

Khaled Koutini, Hamid Eghbal-Zadeh, Gerhard Widmer

Institute of Computational Perception

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings › peer-review

Abstract

In this paper, we present our approach used for the CP-JKU submission in Task 4 of the DCASE-2018 Challenge. We propose a novel iterative knowledge distillation technique for weakly-labeled semi-supervised event detection using neural networks, specifically Recurrent Convolutional Neural Networks (R-CNNs). R-CNNs are used to tag the unlabeled data and predict strong labels. Further, we use the R-CNN strong pseudo-labels on the training datasets and train new models after applying label-smoothing techniques on the strong pseudo-labels. Our proposed approach significantly improved the performance of the baseline, achieving the event-based f-measure of 40.86% compared to 15.11% event-based f-measure of the baseline in the provided test set from the development dataset.

Original language	English
Title of host publication	Proceedings of the Detection and Classification of Acoustic Scenes and Events 2018 Workshop (DCASE2018)
Number of pages	5
Publication status	Published - Nov 2018

Fields of science

202002 Audiovisual media
102 Computer Sciences
102001 Artificial intelligence
102003 Image processing
102015 Information systems

JKU Focus areas

Computation in Informatics and Mathematics
Engineering and Natural Sciences (in general)

Cite this

@inproceedings{89fe5b2822ae40c193eb645734120cd8,

title = "Iterative Knowledge Distillation in R-CNNS for Weakly-Labeled Semi-Supervised Sound Event Detection",

abstract = "In this paper, we present our approach used for the CP-JKU submission in Task 4 of the DCASE-2018 Challenge. We propose a novel iterative knowledge distillation technique for weakly-labeled semi-supervised event detection using neural networks, specifically Recurrent Convolutional Neural Networks (R-CNNs). R-CNNs are used to tag the unlabeled data and predict strong labels. Further, we use the R-CNN strong pseudo-labels on the training datasets and train new models after applying label-smoothing techniques on the strong pseudo-labels. Our proposed approach significantly improved the performance of the baseline, achieving the event-based f-measure of 40.86\% compared to 15.11\% event-based f-measure of the baseline in the provided test set from the development dataset.",

author = "Khaled Koutini and Hamid Eghbal-Zadeh and Gerhard Widmer",

year = "2018",

month = nov,

language = "English",

booktitle = "Proceedings of the Detection and Classification of Acoustic Scenes and Events 2018 Workshop (DCASE2018)",

}

Iterative Knowledge Distillation in R-CNNS for Weakly-Labeled Semi-Supervised Sound Event Detection. / Koutini, Khaled; Eghbal-Zadeh, Hamid; Widmer, Gerhard.
Proceedings of the Detection and Classification of Acoustic Scenes and Events 2018 Workshop (DCASE2018). 2018.

Research output: Chapter in Book/Report/Conference proceeding › Conference proceedings › peer-review

TY - GEN

T1 - Iterative Knowledge Distillation in R-CNNS for Weakly-Labeled Semi-Supervised Sound Event Detection

AU - Koutini, Khaled

AU - Eghbal-Zadeh, Hamid

AU - Widmer, Gerhard

PY - 2018/11

Y1 - 2018/11

N2 - In this paper, we present our approach used for the CP-JKU submission in Task 4 of the DCASE-2018 Challenge. We propose a novel iterative knowledge distillation technique for weakly-labeled semi-supervised event detection using neural networks, specifically Recurrent Convolutional Neural Networks (R-CNNs). R-CNNs are used to tag the unlabeled data and predict strong labels. Further, we use the R-CNN strong pseudo-labels on the training datasets and train new models after applying label-smoothing techniques on the strong pseudo-labels. Our proposed approach significantly improved the performance of the baseline, achieving the event-based f-measure of 40.86% compared to 15.11% event-based f-measure of the baseline in the provided test set from the development dataset.

AB - In this paper, we present our approach used for the CP-JKU submission in Task 4 of the DCASE-2018 Challenge. We propose a novel iterative knowledge distillation technique for weakly-labeled semi-supervised event detection using neural networks, specifically Recurrent Convolutional Neural Networks (R-CNNs). R-CNNs are used to tag the unlabeled data and predict strong labels. Further, we use the R-CNN strong pseudo-labels on the training datasets and train new models after applying label-smoothing techniques on the strong pseudo-labels. Our proposed approach significantly improved the performance of the baseline, achieving the event-based f-measure of 40.86% compared to 15.11% event-based f-measure of the baseline in the provided test set from the development dataset.

M3 - Conference proceedings

BT - Proceedings of the Detection and Classification of Acoustic Scenes and Events 2018 Workshop (DCASE2018)

ER -