期刊
IEEE JOURNAL OF BIOMEDICAL AND HEALTH INFORMATICS
卷 25, 期 10, 页码 3834-3843出版社
IEEE-INST ELECTRICAL ELECTRONICS ENGINEERS INC
DOI: 10.1109/JBHI.2021.3092396
关键词
Convolution; Sensors; Training; Time series analysis; Task analysis; Recycling; Kernel; Sensor; convolutional neural networks; activity recognition; deep learning; channel-selectivity
类别
资金
- National Science Foundation of China [61203237]
- Industry-Academia Cooperation Innovation Fund Projection of Jiangsu Province [BY2016001-02]
- Natural Science Foundation of Jiangsu Province [BK20191371]
This paper explores the use of shallow convolutional neural networks with channel-selectivity for sensor based human activity recognition tasks. Experimental results show that channel-selectivity CNN can achieve lower test errors than traditional static layers, improving the performance of deep HAR without additional costs.
Recently, the state-of-the-art performance in various sensor based human activity recognition (HAR) tasks have been acquired by deep learning, which can extract automatically features from raw data. In order to obtain the best accuracy, many static layers have been always used to train deep neural networks, and their weight connectivity in network remains unchanged. Pursuing the best accuracy in mobile platforms with a very limited computational budget at millions of FLOPs is impractical. In this paper, we make use of shallow convolutional neural networks (CNNs) with channel-selectivity for the use of HAR. As we have known, it is for the first time to adopt channel-selectivity CNN for sensor based HAR tasks. We perform extensive experiments on 5 public benchmark HAR datasets consisting of UCI-HAR dataset, OPPORTUNITY dataset, UniMib-SHAR dataset, WISDM dataset, and PAMAP2 dataset. As a result, the channel-selectivity can achieve lower test errors than static layers. The existing performance of deep HAR can be further improved by the CNN with channel-selectivity without any extra cost.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据