QQ客服 官方微博 反馈留言

	Real-time Activity Recognition on Smartphones Using Deep Neural Networks
	Zhang, Licheng ; Wu, Xihong ; Luo, Dingsheng
	2015
关键词	deep neural networks real-time activity recognition smartphone accelerometer feature learning WEARABLE SENSORS
英文摘要	Real-time activity recognition is an important research problem, which has drawn the attention of many researchers for many years. All previous works of real-time activity recognition needed to manually extract features and then fed these features into the classifiers as the inputs. However, due to the recognition time limit, they could only extract a small amount of features, which might not achieve good accuracy and could be improved. Besides, these features were manually extracted according to the experience of the researchers. Because researchers did not know which features worked better for classification, it was hard to choose suitable features, thus these manually extracted features might not be related to the specific classification task and have poor discrimination ability and could be improved. In this paper, we recommend deep neural networks for real-time activity recognition, which automatically learn suitable features and then perform classification. We collected accelerometer data of seven activities of interest on an android smartphone, including walking, running, standing, sitting, lying, walking upstairs and walking downstairs, and conducted experiments on our collected dataset to compare our method with traditional methods. Besides, we implemented a deep neural network on a smartphone and tested the recognition time of the model. The results showed that deep neural networks could achieve real-time performance and got higher accuracy than traditional methods.; National Basic Research Program (973 Program) of China [2013CB329304]; National Natural Science Foundation of China [90920302, 91120001, 61121002]; "Twelfth Five Year" National Science & Technology Support Program of China [2012BAI12B01]; Key Program of National Social Science Foundation of China [12ZD119]; EI; CPCI-S(ISTP); 1236-1242
语种	英语
出处	12th IEEE Int Conf Ubiquitous Intelligence & Comp/12th IEEE Int Conf Autonom & Trusted Comp/15th IEEE Int Conf Scalable Comp & Commun & Associated Workshops/IEEE Int Conf Cloud & Big Data Comp/IEEE Int Conf Internet People
DOI标识	10.1109/UIC-ATC-ScalCom-CBDCom-IoP.2015.224
内容类型	其他
源URL	[http://ir.pku.edu.cn/handle/20.500.11897/449317]
专题	信息科学技术学院
推荐引用方式 GB/T 7714	Zhang, Licheng,Wu, Xihong,Luo, Dingsheng. Real-time Activity Recognition on Smartphones Using Deep Neural Networks. 2015-01-01.

个性服务

查看访问统计

相关权益政策

暂无数据

收藏/分享

除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。