Asymmetric kernel convolutional neural network for acoustic scenes classification

Chien Yao Wang, Jia Ching Wang, Yu Chi Wu, Pao Chi Chang

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

1 Scopus citations

Abstract

In this Zwork, Zwe propose an Asymmetric Kernel Convolutional Neural NetZwork (AKCNN) for Acoustic Scenes Classification (ASC). Its kernel shape is not the traditional square but asymmetric in Zwidth and height. It also uses Weight Normalization (WN) to accelerate the training process because it can early converge the training loss and accuracy. The best of all, WN can help increase the accuracy of ASC. TUT Acoustic Scenes 2016 Dataset [1] is used for evaluation. The result shoZws that AKCNN achieves accuracy 86.7%. If Zwe rank the score in DCASE2016 ASC Challenge, it shoZws that the system Zwould have a higher score than the 5th place.

Original languageEnglish
Title of host publication2017 IEEE International Symposium on Consumer Electronics, ISCE 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages11-12
Number of pages2
ISBN (Electronic)9781538654330
StatePublished - 2 Jul 2017
Event21st IEEE International Symposium on Consumer Electronics, ISCE 2017 - Kuala Lumpur, Malaysia
Duration: 14 Nov 201715 Nov 2017

Publication series

NameProceedings of the International Symposium on Consumer Electronics, ISCE

Conference

Conference21st IEEE International Symposium on Consumer Electronics, ISCE 2017
Country/TerritoryMalaysia
CityKuala Lumpur
Period14/11/1715/11/17

Keywords

  • Acoustic scenes classification
  • Convolutional neural network
  • Deep learning

Fingerprint

Dive into the research topics of 'Asymmetric kernel convolutional neural network for acoustic scenes classification'. Together they form a unique fingerprint.

Cite this