基于注意力模型的多传感器人类活动识别

doi:10.3969/j.issn.1000-1158.2019.06.04

摘要
图/表
参考文献(27)
相关文章 (15)

全文: PDF (1357 KB) HTML (1 KB)
输出: BibTeX | EndNote (RIS)

摘要深度循环神经网络适用于处理时间序列数据, 然而循环神经网络特征提取能力差, 时间依赖关系挖掘不足。针对此问题, 提出了3种注意力机制和长短时记忆(LSTM)神经网络结合的模型用于人类活动识别问题, 并研究了3种注意力机制在不同数据集上单独及配合使用时对模型精度的影响。对于UCI_HAR数据集, 3种注意力LSTM模型准确率分别为94.13%、95.15%和94.81%,高于LSTM模型识别准确率93.2%。此外, 针对人类活动识别的传感器时间序列数据的标签特点, 提出将时间段分类任务转化为分割任务, 设计了2个基于分割任务的注意力门控循环单元(GRU)神经网络模型, Bahdanau注意力GRU模型在Skoda数据集和机会(Oppor)数据集准确率为84.61%和89.54%, 高于基准UNet模型的70.40%和88.51%。

	服务

	把本文推荐给朋友
	加入我的书架
	加入引用管理器
	E-mail Alert
	RSS
	作者相关文章
	王金甲
	周雅倩
	郝智

关键词 ：计量学, 人类活动识别, 长短时记忆神经网络, 注意力机制, 时间段分类, 分割任务

Abstract：The deep recurrent neural network was suitable for processing time series data. However, the feature extraction ability of traditional recurrent neural network was poor, and the time dependency mining was insufficient. In response to the mentioned problem, three models of attention mechanism and long-term short-term memory (LSTM) neural network were proposed for human activity recognition application problems. The effects of these three mechanisms on the accuracy of the models were studied separately and in combination with different datasets. In the UCI_HAR data set, the accuracy rates of the three attention LSTM models were 94.13%,95.15% and 94.81%, respectively, which were higher than the identification accuracy rate of LSTM model (93.2%). In addition, for the label characteristics of sensor time series data for human activity recognition, it was proposed to convert the time segment classification task into a segmentation task. Therefore, two attention-based gate recurrent unit(GRU) models based on segmentation tasks were designed. The accuracy of Bahdanau attention GRU model were 84.61% and 89.54% in the Skoda data set and the opportunity data set which were higher than the benchmark UNet model’s 70.40% and 88.51%.

Key words： metrology human activity recognition long short time memory neural network attention mechanism;time segment classification segmentation task

收稿日期: 2019-01-07 发布日期: 2019-10-10

PACS:

TB973

基金资助:国家自然科学基金(61473339);河北省青年拔尖人才支持计划(［2013］17)

通讯作者: 王金甲 E-mail: wjj@ysu.edu.cn

作者简介: 王金甲(1978-),河南商丘人,燕山大学教授,博士生导师,主要从事信号处理和模式识别方面的研究。E-mail:wjj@ysu.edu.cn

引用本文:

王金甲,周雅倩,郝智. 基于注意力模型的多传感器人类活动识别[J]. 计量学报, 2019, 40(6): 958-969.
WANG Jin-jia,ZHOU Ya-qian,HAO Zhi. Multi-sensor Human Activity Recognition Based on Attention Model. Acta Metrologica Sinica, 2019, 40(6): 958-969.

链接本文:

http://jlxb.china-csm.org:81/Jwk_jlxb/CN/10.3969/j.issn.1000-1158.2019.06.04 或 http://jlxb.china-csm.org:81/Jwk_jlxb/CN/Y2019/V40/I6/958

［1］Machot F A, Mosa A H, Ali M, et al. Activity Recognition in Sensor Data Streams for Active and Assisted Living Environments［J］. IEEE Transactions on Circuits & Systems for Video Technology, 2018, 28(10): 2933-2945.
［2］Lu L, Cai Q L, Zhan Y J . Activity Recognition in Smart Homes［J］. Multimedia Tools and Applications, 2017, 76(22): 24203-24220.
［3］罗晓宇. 基于DBN-HMM的人体动作识别［D］. 西安:西安理工大学, 2018.
［4］Wen J H, Wang Z Y. Learning general model for activity recognition with limited labelled data［J］. Expert Systems with Applications, 2017, 74(5): 19-28.
［5］陈文超. 基于条件随机场的人体动作识别［D］. 成都:电子科技大学, 2016.
［6］Nweke H F, Teh Y W, Al-Garadi M A, et al. Deep Learning Algorithms for Human Activity Recognition using Mobile and Wearable Sensor Networks: State of the Art and Research Challenges［J］. Expert Systems with Applications, 2018, 105(9):233-261.
［7］Ronao C A, Cho S B. Human activity recognition with smartphone sensors using deep learning neural networks［J］. Expert Systems with Applications, 2016, 59(10):235-244.
［8］Yang J, Nguyen M N, San P P, et al. Deep Convolutional Neural Networks on Multichannel Time Series for Human Activity Recognition［C］//IJCAI. IJCAI’15 Proceedings of the 24th International Conference on Artificial Intelligence. Buenos Aires, Argentina, 2015: 3995-4001.
［9］李艇. 基于3D CNN的人体连续动作识别［D］. 哈尔滨: 哈尔滨工业大学, 2018.
［10］Ordóez F J, Roggen D. Deep convolutional and lstm recurrent neural networks for multimodal wearable activity recognition［J］. Sensors, 2016, 16(1): 115.
［11］Arifoglu D, Bouchachia A. Activity recognition and abnormal behaviour detection with recurrent neural networks［J］. Procedia Computer Science, 2017, 110: 86-93.
［12］张儒鹏, 于亚新, 张康, 等. 基于 OI-LSTM 神经网络结构的人类动作识别模型研究［J］. 计算机科学与探索, 2018, 12(12): 1926-1939.
Zhang R P, Yu Y X, Zhang K, et al. Research on human action recognition model based on OI-LSTM neural network structure［J］. Journal of Frontiers of Computer Science and Technology, 2018, 12(12): 1926-1939.

［13］Liu J, Wang G, Hu P, et al. Global Context-Aware Attention LSTM Networks for 3D Action Recognition［C］//IEEE Computer Society. 2017 IEEE Conference on Computer Vision and Pattern Recognition(CVPR).
Hawaii, USA, 2017:1647-1656.
［14］Du W, Wang Y, Qiao Y. Recurrent Spatial-Temporal Attention Network for Action Recognition in Videos［J］. IEEE Transactions on Image Processing, 2018, 27(3): 1347-1360.
［15］宇文浩. 基于注意机制的卷积递归神经网络的动作识别［D］. 大连:大连理工大学, 2017.
［16］Ronneberger O, Fischer P, Brox T. U-net: Convolutional networks for biomedical image segmentation［C］//TUM (Munich) and FAU (Erlangen-Nuremberg). The 18th International Conference on Medical Image Computing and Computer Assisted Intervention, Munich, Germany. 2015: 234-241.
［17］Hongyi Z, Hsinchun C, Randall B. A sequence-to-sequence model-based deep learning approach for recognizing activity of daily living for senior care［J］. Journal of Biomedical Informatics, 2018, 84(8): 148-158.
［18］Wu L, Tian F, Zhao L, et al. Word attention for sequence to sequence text understanding［C］//AAAI Association. Thirty-Second AAAI Conference on Artificial Intelligence. New Orleans, USA,2018: 5578-5585.
［19］Kadari R, Zhang Y, Zhang W, et al. CCG supertagging via Bidirectional LSTM-CRF neural architecture［J］. Neurocomputing, 2018, 283(3): 31-37.
［20］Ergen T, Kozat S S. Efficient Online Learning Algorithms Based on LSTM Neural Networks［J］. IEEE Transactions on Neural Networks and Learning Systems, 2017,29(8):3772-3783.
［21］Anguita D, Ghio A, Oneto L, et al. A public domain dataset for human activity recognition using smartphones［C］//ESANN. European Symposium on Artificial Neural Networks. Bruges, Belgium, 2013: 11-14.
［22］Xu Y, Shen Z, Zhang X, et al. Learning multi-level features for sensor-based human action recognition［J］. Pervasive and Mobile Computing, 2017, 40: 324-338.
［23］Chavarriaga R, Sagha H, Calatroni A, et al. The Opportunity challenge: A benchmark database for on-body sensor-based activity recognition［J］. Pattern Recognition Letters, 2013, 34(15): 2033-2042.
［24］Wang A. A Comparative Study on Human Activity Recognition Using Inertial Sensors in a Smartphone［J］. IEEE Sensors Journal, 2016, 16(11): 4566-4578.
［25］Cao L, Wang Y, Zhang B, et al. GCHAR: An efficient Group-based Context—aware human activity recognition on smartphone［J］. Journal of Parallel and Distributed Computing, 2018, 118: 67-80.
［26］Zeng M, Nguyen L T, Yu B, et al. Convolutional Neural Networks for human activity recognition using mobile sensors［C］// IEEE. Sixth International Conference on Mobile Computing. Austin. TX, USA,2014: 1-18.
［27］Alsheikh M A, Selim A, Niyato D, et al. Deep Activity Recognition Models with Triaxial Accelerometers［C］//AAAI. AAAI Workshop: Artificial Intelligence Applied to Assistive Technologies and Smart Environments. Phoenix, Arizona, USA, 2016: 8-13.