Fusion Method of Convolutional Neural Network and Support Vector Machine for High Accuracy Anomaly Detection

Fusaomi Nagata, Kenta Tokuno, Kento Nakashima, Akimasa Otsuka, Takeshi Ikeda, Hiroyuki Ochi, Keigo Watanabe, Maki K. Habib

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

Abstract

In this paper, binary classification methods using support vector machines (SVM) obtained from one-class learning and two-class learning are introduced. Firstly, an application of deep CNN (DCNN) for visual inspection is developed and is trained using a large number of images to inspect undesirable defects seen in resin molded articles. Then, the trained DCNN named sssNet and well-known Alexnet are respectively incorporated with two kinds of one-class learning based SVMs to classify sample images with high recognition rate into accept as OK category or reject as NG category, in which compressed feature vectors obtained from the DCNNs are used as the inputs for the SVMs. The performances of two SVMs obtained from one-class learning are compared and evaluated through training and classification experiments. Then, another SVM obtained from two-class learning is introduced. Finally, a template matching technique is further applied to extract important target areas from original training and test images. This will be able to enhance the reliability and accuracy for binary classification.

Original languageEnglish
Title of host publicationProceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages970-975
Number of pages6
ISBN (Electronic)9781728116983
DOIs
Publication statusPublished - Aug 1 2019
Event16th IEEE International Conference on Mechatronics and Automation, ICMA 2019 - Tianjin, China
Duration: Aug 4 2019Aug 7 2019

Publication series

NameProceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019

Conference

Conference16th IEEE International Conference on Mechatronics and Automation, ICMA 2019
CountryChina
CityTianjin
Period8/4/198/7/19

Fingerprint

Anomaly Detection
Support vector machines
Support Vector Machine
Fusion
High Accuracy
Fusion reactions
Neural Networks
Neural networks
Binary Classification
Template matching
Resins
Inspection
Template Matching
Defects
Feature Vector
Classify
Class
Learning
Experiments
Target

Keywords

  • Deep Convolutional Neural Network (DCNN)
  • Defect Inspection System
  • Support Vector Machine (SVM)

ASJC Scopus subject areas

  • Signal Processing
  • Mechanical Engineering
  • Control and Optimization
  • Artificial Intelligence

Cite this

Nagata, F., Tokuno, K., Nakashima, K., Otsuka, A., Ikeda, T., Ochi, H., ... Habib, M. K. (2019). Fusion Method of Convolutional Neural Network and Support Vector Machine for High Accuracy Anomaly Detection. In Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019 (pp. 970-975). [8816454] (Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/ICMA.2019.8816454

Fusion Method of Convolutional Neural Network and Support Vector Machine for High Accuracy Anomaly Detection. / Nagata, Fusaomi; Tokuno, Kenta; Nakashima, Kento; Otsuka, Akimasa; Ikeda, Takeshi; Ochi, Hiroyuki; Watanabe, Keigo; Habib, Maki K.

Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019. Institute of Electrical and Electronics Engineers Inc., 2019. p. 970-975 8816454 (Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019).

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

Nagata, F, Tokuno, K, Nakashima, K, Otsuka, A, Ikeda, T, Ochi, H, Watanabe, K & Habib, MK 2019, Fusion Method of Convolutional Neural Network and Support Vector Machine for High Accuracy Anomaly Detection. in Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019., 8816454, Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019, Institute of Electrical and Electronics Engineers Inc., pp. 970-975, 16th IEEE International Conference on Mechatronics and Automation, ICMA 2019, Tianjin, China, 8/4/19. https://doi.org/10.1109/ICMA.2019.8816454
Nagata F, Tokuno K, Nakashima K, Otsuka A, Ikeda T, Ochi H et al. Fusion Method of Convolutional Neural Network and Support Vector Machine for High Accuracy Anomaly Detection. In Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019. Institute of Electrical and Electronics Engineers Inc. 2019. p. 970-975. 8816454. (Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019). https://doi.org/10.1109/ICMA.2019.8816454
Nagata, Fusaomi ; Tokuno, Kenta ; Nakashima, Kento ; Otsuka, Akimasa ; Ikeda, Takeshi ; Ochi, Hiroyuki ; Watanabe, Keigo ; Habib, Maki K. / Fusion Method of Convolutional Neural Network and Support Vector Machine for High Accuracy Anomaly Detection. Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019. Institute of Electrical and Electronics Engineers Inc., 2019. pp. 970-975 (Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019).
@inproceedings{6f1d1f593b984ba980b406de3ad885f7,
title = "Fusion Method of Convolutional Neural Network and Support Vector Machine for High Accuracy Anomaly Detection",
abstract = "In this paper, binary classification methods using support vector machines (SVM) obtained from one-class learning and two-class learning are introduced. Firstly, an application of deep CNN (DCNN) for visual inspection is developed and is trained using a large number of images to inspect undesirable defects seen in resin molded articles. Then, the trained DCNN named sssNet and well-known Alexnet are respectively incorporated with two kinds of one-class learning based SVMs to classify sample images with high recognition rate into accept as OK category or reject as NG category, in which compressed feature vectors obtained from the DCNNs are used as the inputs for the SVMs. The performances of two SVMs obtained from one-class learning are compared and evaluated through training and classification experiments. Then, another SVM obtained from two-class learning is introduced. Finally, a template matching technique is further applied to extract important target areas from original training and test images. This will be able to enhance the reliability and accuracy for binary classification.",
keywords = "Deep Convolutional Neural Network (DCNN), Defect Inspection System, Support Vector Machine (SVM)",
author = "Fusaomi Nagata and Kenta Tokuno and Kento Nakashima and Akimasa Otsuka and Takeshi Ikeda and Hiroyuki Ochi and Keigo Watanabe and Habib, {Maki K.}",
year = "2019",
month = "8",
day = "1",
doi = "10.1109/ICMA.2019.8816454",
language = "English",
series = "Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "970--975",
booktitle = "Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019",

}

TY - GEN

T1 - Fusion Method of Convolutional Neural Network and Support Vector Machine for High Accuracy Anomaly Detection

AU - Nagata, Fusaomi

AU - Tokuno, Kenta

AU - Nakashima, Kento

AU - Otsuka, Akimasa

AU - Ikeda, Takeshi

AU - Ochi, Hiroyuki

AU - Watanabe, Keigo

AU - Habib, Maki K.

PY - 2019/8/1

Y1 - 2019/8/1

N2 - In this paper, binary classification methods using support vector machines (SVM) obtained from one-class learning and two-class learning are introduced. Firstly, an application of deep CNN (DCNN) for visual inspection is developed and is trained using a large number of images to inspect undesirable defects seen in resin molded articles. Then, the trained DCNN named sssNet and well-known Alexnet are respectively incorporated with two kinds of one-class learning based SVMs to classify sample images with high recognition rate into accept as OK category or reject as NG category, in which compressed feature vectors obtained from the DCNNs are used as the inputs for the SVMs. The performances of two SVMs obtained from one-class learning are compared and evaluated through training and classification experiments. Then, another SVM obtained from two-class learning is introduced. Finally, a template matching technique is further applied to extract important target areas from original training and test images. This will be able to enhance the reliability and accuracy for binary classification.

AB - In this paper, binary classification methods using support vector machines (SVM) obtained from one-class learning and two-class learning are introduced. Firstly, an application of deep CNN (DCNN) for visual inspection is developed and is trained using a large number of images to inspect undesirable defects seen in resin molded articles. Then, the trained DCNN named sssNet and well-known Alexnet are respectively incorporated with two kinds of one-class learning based SVMs to classify sample images with high recognition rate into accept as OK category or reject as NG category, in which compressed feature vectors obtained from the DCNNs are used as the inputs for the SVMs. The performances of two SVMs obtained from one-class learning are compared and evaluated through training and classification experiments. Then, another SVM obtained from two-class learning is introduced. Finally, a template matching technique is further applied to extract important target areas from original training and test images. This will be able to enhance the reliability and accuracy for binary classification.

KW - Deep Convolutional Neural Network (DCNN)

KW - Defect Inspection System

KW - Support Vector Machine (SVM)

UR - http://www.scopus.com/inward/record.url?scp=85072405511&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85072405511&partnerID=8YFLogxK

U2 - 10.1109/ICMA.2019.8816454

DO - 10.1109/ICMA.2019.8816454

M3 - Conference contribution

AN - SCOPUS:85072405511

T3 - Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019

SP - 970

EP - 975

BT - Proceedings of 2019 IEEE International Conference on Mechatronics and Automation, ICMA 2019

PB - Institute of Electrical and Electronics Engineers Inc.

ER -