东北大学学报(自然科学版) ›› 2022, Vol. 43 ›› Issue (6): 776-782.DOI: 10.12068/j.issn.1005-3026.2022.06.003

• 信息与控制 • 上一篇    下一篇

一种应用于人体活动识别的迁移学习算法

赵海, 陈佳伟, 施瀚, 王相   

  1. (东北大学 计算机科学与工程学院, 辽宁 沈阳110169)
  • 修回日期:2021-07-29 接受日期:2021-07-29 发布日期:2022-07-01
  • 通讯作者: 赵海
  • 作者简介:赵海(1959 -),男,辽宁沈阳人,东北大学教授,博士生导师.
  • 基金资助:
    国家重点研发计划项目(2019JSJ12ZDYF01); 中央高校基本科研业务费专项资金资助项目(2020GFZD014).

A Transfer Learning Algorithm Applied to Human Activity Recognition

ZHAO Hai, CHEN Jia-wei, SHI Han, WANG Xiang   

  1. School of Computer Science & Engineering, Northeastern University, Shenyang 110169, China.
  • Revised:2021-07-29 Accepted:2021-07-29 Published:2022-07-01
  • Contact: CHEN Jia-wei
  • About author:-
  • Supported by:
    -

摘要: 通过采集可穿戴运动传感器信号,并利用迁移学习克服数据分布不一致来识别人体日常行为成为当下主流.利用可穿戴传感器采集信号,会产生影响迁移效果的噪声样本,传统的算法缺少对这部分样本的处理.针对这一问题,在传统算法的基础上进行改进,引入了基于马氏距离的样本筛选算法,提出了可用于人体活动识别的迁移学习算法T-WMD,并在两个公开的人体活动识别数据集上与其他5种算法进行对比实验.结果表明提出的算法可以有效地提升迁移学习效果.

关键词: 生理信号;人体活动识别;迁移学习;体域网;机器学习

Abstract: Collecting wearable motion sensor signals and using transfer learning to overcome the inconsistency of data distribution to identify the daily behavior of the human body are very popular technologies. Using wearable sensors to collect signals will result in generating noise samples that affect the transfer effect. Traditional algorithms lack the processing of these samples. To solve this problem, the traditional algorithm was improved by introducing a sample screening algorithm based on Mahalanobis distance, and a transfer learning algorithm T-WMD was proposed that can be used for human activity recognition. And compared with other five algorithms on two public human activity recognition data sets, the results show that the algorithm proposed in this paper can effectively improve the effect of transfer learning.

Key words: physiological signal; human activity recognition; transfer learning; body area network; machine learning

中图分类号: