Acoustic Scene Classification with Reject Option Based on Resnets

Bernhard Lehner; Khaled Koutini; Christopher Hartwig Schwarzlmüller; Thomas Gallien; Gerhard Widmer

Acoustic Scene Classification with Reject Option Based on Resnets

Bernhard Lehner, Khaled Koutini, Christopher Hartwig Schwarzlmüller, Thomas Gallien, Gerhard Widmer

Embedded AI (eAI)

Research output: Working paper

Abstract

This technical report describes the submissions from the SAL/CP JKU team for Task 1 - Subtask C (classification on data that includes classes not encountered in the training data) of the DCASE-2019 challenge. Our method uses a ResNet variant specifically adapted to be used along with spectrograms in the context of Acoustic Scene Classification (ASC). The reject option is based on the logit values of the same networks. We do not use any of the provided external data sets, and perform data augmentation only with the mixup technique [1]. The result of our experiments is a system that achieves classification accuracies of up to around 60Leaderboard. This is an improvement of around 14 percentage points compared to the official DCASE 2019 baseline.

Original language	American English
Publication status	Published - 1 Jun 2019

Cite this

@techreport{ee2bcb2eff224534bf996bfee8d66a11,

title = "Acoustic Scene Classification with Reject Option Based on Resnets",

abstract = "This technical report describes the submissions from the SAL/CP JKU team for Task 1 - Subtask C (classification on data that includes classes not encountered in the training data) of the DCASE-2019 challenge. Our method uses a ResNet variant specifically adapted to be used along with spectrograms in the context of Acoustic Scene Classification (ASC). The reject option is based on the logit values of the same networks. We do not use any of the provided external data sets, and perform data augmentation only with the mixup technique [1]. The result of our experiments is a system that achieves classification accuracies of up to around 60Leaderboard. This is an improvement of around 14 percentage points compared to the official DCASE 2019 baseline.",

author = "Bernhard Lehner and Khaled Koutini and Schwarzlm{\"u}ller, {Christopher Hartwig} and Thomas Gallien and Gerhard Widmer",

year = "2019",

month = jun,

day = "1",

language = "Englisch (Amerika)",

type = "WorkingPaper",

}

TY - UNPB

T1 - Acoustic Scene Classification with Reject Option Based on Resnets

AU - Lehner, Bernhard

AU - Koutini, Khaled

AU - Schwarzlmüller, Christopher Hartwig

AU - Gallien, Thomas

AU - Widmer, Gerhard

PY - 2019/6/1

Y1 - 2019/6/1

N2 - This technical report describes the submissions from the SAL/CP JKU team for Task 1 - Subtask C (classification on data that includes classes not encountered in the training data) of the DCASE-2019 challenge. Our method uses a ResNet variant specifically adapted to be used along with spectrograms in the context of Acoustic Scene Classification (ASC). The reject option is based on the logit values of the same networks. We do not use any of the provided external data sets, and perform data augmentation only with the mixup technique [1]. The result of our experiments is a system that achieves classification accuracies of up to around 60Leaderboard. This is an improvement of around 14 percentage points compared to the official DCASE 2019 baseline.

AB - This technical report describes the submissions from the SAL/CP JKU team for Task 1 - Subtask C (classification on data that includes classes not encountered in the training data) of the DCASE-2019 challenge. Our method uses a ResNet variant specifically adapted to be used along with spectrograms in the context of Acoustic Scene Classification (ASC). The reject option is based on the logit values of the same networks. We do not use any of the provided external data sets, and perform data augmentation only with the mixup technique [1]. The result of our experiments is a system that achieves classification accuracies of up to around 60Leaderboard. This is an improvement of around 14 percentage points compared to the official DCASE 2019 baseline.

UR - http://dcase.community/challenge2019/task-acoustic-scene-classification-results-c

M3 - Arbeitspapier

BT - Acoustic Scene Classification with Reject Option Based on Resnets

ER -

Acoustic Scene Classification with Reject Option Based on Resnets

Abstract

Other files and links

Cite this