文献类型：期刊文献

英文题名：Research on Classification Algorithm Based on Multivariate Time Series

作者：Wan, Siyuan[1]; Chen, Tianyu[1]; Ni, Xue[1]; Xu, Chunyu[1]; Wang, Rong[1]; Wan, Yongjing[1]

机构：[1] School of Information Science and Engineering, East China University of Science and Technology, Shanghai, Shanghai, 200237, China

年份：2021

起止页码：298

外文期刊名：ACM International Conference Proceeding Series

收录：EI(收录号：20221111792213)

语种：英文

外文关键词：Convolution - Time series

摘要：In the last decade, multivariate time series classification has gotten a lot of interest as a research hotspot. In this paper, an existing univariate time series classification model is combined with a long-period full convolutional network (LSTM-FCN-CBAM) from the Convolutional Block Attention Module (CBAM, Convolutional Block Attention Module), resulting in a multivariate time series classification model with improved classification accuracy. This model outperforms most advanced models while requiring very little preprocessing. This model is capable of handling a wide range of difficult and multi-time series classification problems, including activity and behavior detection. Furthermore, the model is more efficient and smaller in size, making it ideal for usage in memoryconstrained systems. ? 2021 ACM.