Paper Abstract and Keywords |
Presentation |
2021-03-05 16:55
Electromagnetic Noise Classification and Novelty Detection for Frequency Sharing among Various Communications in the Manufacturing Field Michio Miyamoto, Ayano Ohnishi, Yoshio Takeuchi (ATR), Toshiyuki Maeyama (ATR/Takushoku Univ.), Akio Hasegawa, Hiroyuki Yokoyama (ATR) SR2020-89 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
Machine learning is applied to classify the electromagnetic noise generated in the manufacturing field. The frequency and pattern of electromagnetic noise generated in the manufacturing field depends on the source equipment. Therefore, there is a problem that a new noise pattern that occurs infrequently and has not been learned is erroneously classified as a known pattern. Machine learning novelty detection can be used to find unknown patterns. However, when operating measurement, many computer resources are required to operate classification and novelty detection in parallel. In this report, we describe a method for detecting unlearned patterns using the predict probability of classification estimation results by supervised machine learning. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Factory Wireless Communication / Electromagnetic Noise / Machine Learning Classification / Novelty Detection / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 120, no. 405, SR2020-89, pp. 99-103, March 2021. |
Paper # |
SR2020-89 |
Date of Issue |
2021-02-24 (SR) |
ISSN |
Online edition: ISSN 2432-6380 |
Copyright and reproduction |
All rights are reserved and no part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Notwithstanding, instructors are permitted to photocopy isolated articles for noncommercial classroom use without fee. (License No.: 10GA0019/12GB0052/13GB0056/17GB0034/18GB0034) |
Download PDF |
SR2020-89 |