Unsupervised learning for human activity recognition using smartphone sensors
作者：
Highlights：
• We investigate activity recognition using unsupervised learning, with a smartphone.
• The number of activities can be determined by the Caliński–Harabasz index.
• The mixture of Gaussian outperforms when the number of activities is known.
• The hierarchical clustering and DBSCAN attain above 90% accuracy for appropriate settings.
• The study provides an idea for activity recognition methods without training datasets.
摘要
•We investigate activity recognition using unsupervised learning, with a smartphone.•The number of activities can be determined by the Caliński–Harabasz index.•The mixture of Gaussian outperforms when the number of activities is known.•The hierarchical clustering and DBSCAN attain above 90% accuracy for appropriate settings.•The study provides an idea for activity recognition methods without training datasets.
论文关键词：Human activity recognition,Unsupervised learning,Healthcare services,Smartphone sensors,Sensor data analysis
论文评审过程：Available online 6 May 2014.
论文官网地址：https://doi.org/10.1016/j.eswa.2014.04.037
