Human activity data discovery from triaxial accelerometer sensor: Non-supervised learning sensitivity to feature extraction parametrization
作者：
Highlights：
• The presented study performs two different tests in intra and inter subject context.
• A set of 180 features is implemented to be selected based on clustering performance.
• Our algorithm searches for the best feature extraction parameter.
• A new clustering metric based on the construction of the confusion matrix is proposed.
• A novel gesture recognition system based on data from a single 3 dimensional accelerometer.
摘要
•The presented study performs two different tests in intra and inter subject context.•A set of 180 features is implemented to be selected based on clustering performance.•Our algorithm searches for the best feature extraction parameter.•A new clustering metric based on the construction of the confusion matrix is proposed.•A novel gesture recognition system based on data from a single 3 dimensional accelerometer.
论文关键词：Human activity recognition,Interactive knowledge discovery,Feature extraction,Dimensionality reduction,Clustering algorithms
论文评审过程：Received 10 December 2013, Revised 30 June 2014, Accepted 24 July 2014, Available online 15 November 2014.
论文官网地址：https://doi.org/10.1016/j.ipm.2014.07.008
