Paper Abstract and Keywords |
Presentation |
2019-06-17 17:00
Adaptive Discretization based Predictive Sequence Mining for Continuous Time Series Yoshikazu Shibahara, Takuto Sakuma (NIT), Ichiro Takeuchi (NIT/RIKEN/NIMS), Masayuki Karasuyama (NIT/NIMS) IBISML2019-9 |
Abstract |
(in Japanese) |
(See Japanese page) |
(in English) |
In recent years, improvement of sensor performance and spread of portable devices such as smartphones enable us to easily collect time-series data.
Thus, it is an important task to extract valuable information from time series-data.
In this research, we propose a method extracting a class specific patterns from time-series data by using an adaptive discretization algorithm for a continuous feature space.
Conventional approaches need to define a symbolized representation of the original continuous time-series data beforehand.
Our approach can construct a sparse linear model by selecting important patterns from a variety of possible symbolizations.
The proposed method efficiently deals with a huge number of patterns by combining a safe-screening technique and sequence pattern mining.
Our numerical experiments demonstrate effectiveness of our approach through several benchmark datasets. |
Keyword |
(in Japanese) |
(See Japanese page) |
(in English) |
Continuous valued sequence data / sparse modeling / sequence mining / / / / / |
Reference Info. |
IEICE Tech. Rep., vol. 119, no. 89, IBISML2019-9, pp. 57-64, June 2019. |
Paper # |
IBISML2019-9 |
Date of Issue |
2019-06-10 (IBISML) |
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 |
IBISML2019-9 |