Hybrid deep learning based on GAN for classifying BSR noises from invehicle sensors

Jin Young Kim, Seok Jun Bu, Sung Bae Cho

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

BSR (Buzz, squeak, and rattle) noises are essential criteria for the quality of a vehicle. It is necessary to classify them to handle them appropriately. Although many studies have been conducted to classify noise, they suffered some problems: the difficulty in extracting features, a small amount of data to train a classifier, and less robustness to background noise. This paper proposes a method called transferred encoder-decoder generative adversarial networks (tedGAN) which solves the problems. Deep auto-encoder (DAE) compresses and reconstructs the audio data for capturing the features of them. The decoder network is transferred to the generator of GAN so as to make the process of training generator more stable. Because the generator and the discriminator of GAN are trained at the same time, the capacity of extracting features is enhanced, and a knowledge space of the data is expanded with a small amount of data. The discriminator to classify whether the input is the real or fake BSR noises is transferred again to the classifier; then it is finally trained to classify the BSR noises. The classifier yields the accuracy of 95.15%, which outperforms other machine learning models. We analyze the model with t-SNE algorithm to investigate the misclassified data. The proposed model achieves the accuracy of 92.05% for the data including background noise.

Original languageEnglish
Title of host publicationHybrid Artificial Intelligent Systems - 13th International Conference, HAIS 2018, Proceedings
EditorsAlvaro Herrero, Hector Quintian, Jose Antonio Saez, Emilio Corchado, Francisco Javier de Cos Juez, Jose Ramon Villar, Enrique A. de la Cal
PublisherSpringer Verlag
Pages27-38
Number of pages12
ISBN (Print)9783319926384
DOIs
Publication statusPublished - 2018 Jan 1
Event13th International Conference on Hybrid Artificial Intelligent Systems, HAIS 2018 - Oviedo, Spain
Duration: 2018 Jun 202018 Jun 22

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Volume10870 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Other

Other13th International Conference on Hybrid Artificial Intelligent Systems, HAIS 2018
CountrySpain
CityOviedo
Period18/6/2018/6/22

Fingerprint

Hybrid Learning
Classifiers
Discriminators
Sensor
Sensors
Classify
Classifier
Generator
Encoder
Learning systems
Deep learning
Machine Learning
Model
Robustness
Necessary

All Science Journal Classification (ASJC) codes

  • Theoretical Computer Science
  • Computer Science(all)

Cite this

Kim, J. Y., Bu, S. J., & Cho, S. B. (2018). Hybrid deep learning based on GAN for classifying BSR noises from invehicle sensors. In A. Herrero, H. Quintian, J. Antonio Saez, E. Corchado, F. J. de Cos Juez, J. R. Villar, & E. A. de la Cal (Eds.), Hybrid Artificial Intelligent Systems - 13th International Conference, HAIS 2018, Proceedings (pp. 27-38). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10870 LNAI). Springer Verlag. https://doi.org/10.1007/978-3-319-92639-1_3
Kim, Jin Young ; Bu, Seok Jun ; Cho, Sung Bae. / Hybrid deep learning based on GAN for classifying BSR noises from invehicle sensors. Hybrid Artificial Intelligent Systems - 13th International Conference, HAIS 2018, Proceedings. editor / Alvaro Herrero ; Hector Quintian ; Jose Antonio Saez ; Emilio Corchado ; Francisco Javier de Cos Juez ; Jose Ramon Villar ; Enrique A. de la Cal. Springer Verlag, 2018. pp. 27-38 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).
@inproceedings{5c82d6172330449eaab2ad1e2d009212,
title = "Hybrid deep learning based on GAN for classifying BSR noises from invehicle sensors",
abstract = "BSR (Buzz, squeak, and rattle) noises are essential criteria for the quality of a vehicle. It is necessary to classify them to handle them appropriately. Although many studies have been conducted to classify noise, they suffered some problems: the difficulty in extracting features, a small amount of data to train a classifier, and less robustness to background noise. This paper proposes a method called transferred encoder-decoder generative adversarial networks (tedGAN) which solves the problems. Deep auto-encoder (DAE) compresses and reconstructs the audio data for capturing the features of them. The decoder network is transferred to the generator of GAN so as to make the process of training generator more stable. Because the generator and the discriminator of GAN are trained at the same time, the capacity of extracting features is enhanced, and a knowledge space of the data is expanded with a small amount of data. The discriminator to classify whether the input is the real or fake BSR noises is transferred again to the classifier; then it is finally trained to classify the BSR noises. The classifier yields the accuracy of 95.15{\%}, which outperforms other machine learning models. We analyze the model with t-SNE algorithm to investigate the misclassified data. The proposed model achieves the accuracy of 92.05{\%} for the data including background noise.",
author = "Kim, {Jin Young} and Bu, {Seok Jun} and Cho, {Sung Bae}",
year = "2018",
month = "1",
day = "1",
doi = "10.1007/978-3-319-92639-1_3",
language = "English",
isbn = "9783319926384",
series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",
publisher = "Springer Verlag",
pages = "27--38",
editor = "Alvaro Herrero and Hector Quintian and {Antonio Saez}, Jose and Emilio Corchado and {de Cos Juez}, {Francisco Javier} and Villar, {Jose Ramon} and {de la Cal}, {Enrique A.}",
booktitle = "Hybrid Artificial Intelligent Systems - 13th International Conference, HAIS 2018, Proceedings",
address = "Germany",

}

Kim, JY, Bu, SJ & Cho, SB 2018, Hybrid deep learning based on GAN for classifying BSR noises from invehicle sensors. in A Herrero, H Quintian, J Antonio Saez, E Corchado, FJ de Cos Juez, JR Villar & EA de la Cal (eds), Hybrid Artificial Intelligent Systems - 13th International Conference, HAIS 2018, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), vol. 10870 LNAI, Springer Verlag, pp. 27-38, 13th International Conference on Hybrid Artificial Intelligent Systems, HAIS 2018, Oviedo, Spain, 18/6/20. https://doi.org/10.1007/978-3-319-92639-1_3

Hybrid deep learning based on GAN for classifying BSR noises from invehicle sensors. / Kim, Jin Young; Bu, Seok Jun; Cho, Sung Bae.

Hybrid Artificial Intelligent Systems - 13th International Conference, HAIS 2018, Proceedings. ed. / Alvaro Herrero; Hector Quintian; Jose Antonio Saez; Emilio Corchado; Francisco Javier de Cos Juez; Jose Ramon Villar; Enrique A. de la Cal. Springer Verlag, 2018. p. 27-38 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Vol. 10870 LNAI).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Hybrid deep learning based on GAN for classifying BSR noises from invehicle sensors

AU - Kim, Jin Young

AU - Bu, Seok Jun

AU - Cho, Sung Bae

PY - 2018/1/1

Y1 - 2018/1/1

N2 - BSR (Buzz, squeak, and rattle) noises are essential criteria for the quality of a vehicle. It is necessary to classify them to handle them appropriately. Although many studies have been conducted to classify noise, they suffered some problems: the difficulty in extracting features, a small amount of data to train a classifier, and less robustness to background noise. This paper proposes a method called transferred encoder-decoder generative adversarial networks (tedGAN) which solves the problems. Deep auto-encoder (DAE) compresses and reconstructs the audio data for capturing the features of them. The decoder network is transferred to the generator of GAN so as to make the process of training generator more stable. Because the generator and the discriminator of GAN are trained at the same time, the capacity of extracting features is enhanced, and a knowledge space of the data is expanded with a small amount of data. The discriminator to classify whether the input is the real or fake BSR noises is transferred again to the classifier; then it is finally trained to classify the BSR noises. The classifier yields the accuracy of 95.15%, which outperforms other machine learning models. We analyze the model with t-SNE algorithm to investigate the misclassified data. The proposed model achieves the accuracy of 92.05% for the data including background noise.

AB - BSR (Buzz, squeak, and rattle) noises are essential criteria for the quality of a vehicle. It is necessary to classify them to handle them appropriately. Although many studies have been conducted to classify noise, they suffered some problems: the difficulty in extracting features, a small amount of data to train a classifier, and less robustness to background noise. This paper proposes a method called transferred encoder-decoder generative adversarial networks (tedGAN) which solves the problems. Deep auto-encoder (DAE) compresses and reconstructs the audio data for capturing the features of them. The decoder network is transferred to the generator of GAN so as to make the process of training generator more stable. Because the generator and the discriminator of GAN are trained at the same time, the capacity of extracting features is enhanced, and a knowledge space of the data is expanded with a small amount of data. The discriminator to classify whether the input is the real or fake BSR noises is transferred again to the classifier; then it is finally trained to classify the BSR noises. The classifier yields the accuracy of 95.15%, which outperforms other machine learning models. We analyze the model with t-SNE algorithm to investigate the misclassified data. The proposed model achieves the accuracy of 92.05% for the data including background noise.

UR - http://www.scopus.com/inward/record.url?scp=85048865061&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85048865061&partnerID=8YFLogxK

U2 - 10.1007/978-3-319-92639-1_3

DO - 10.1007/978-3-319-92639-1_3

M3 - Conference contribution

AN - SCOPUS:85048865061

SN - 9783319926384

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 27

EP - 38

BT - Hybrid Artificial Intelligent Systems - 13th International Conference, HAIS 2018, Proceedings

A2 - Herrero, Alvaro

A2 - Quintian, Hector

A2 - Antonio Saez, Jose

A2 - Corchado, Emilio

A2 - de Cos Juez, Francisco Javier

A2 - Villar, Jose Ramon

A2 - de la Cal, Enrique A.

PB - Springer Verlag

ER -

Kim JY, Bu SJ, Cho SB. Hybrid deep learning based on GAN for classifying BSR noises from invehicle sensors. In Herrero A, Quintian H, Antonio Saez J, Corchado E, de Cos Juez FJ, Villar JR, de la Cal EA, editors, Hybrid Artificial Intelligent Systems - 13th International Conference, HAIS 2018, Proceedings. Springer Verlag. 2018. p. 27-38. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). https://doi.org/10.1007/978-3-319-92639-1_3