- Machine Learning Methods for Sensor-based Human Train Recognition with Data Heterogeneity — A Evaluation(arXiv)
Author : : Xiaozhou Ye, Kouichi Sakurai, Nirmal Nair, Kevin I-Kai Wang
Abstract : Sensor-based Human Train Recognition (HAR) is crucial in ubiquitous computing, analysing behaviours by the use of multi-dimensional observations. No matter evaluation progress, HAR confronts challenges, considerably in information distribution assumptions. Most analysis often assume uniform information distributions all through datasets, contrasting with the varied nature of wise sensor information in human actions. Addressing information heterogeneity factors can improve effectivity, reduce computational costs, and assist in rising personalised, adaptive fashions with a lot much less annotated information. This overview investigates how machine learning addresses information heterogeneity in HAR, by categorizing information heterogeneity varieties, making use of corresponding applicable machine learning methods, summarizing obtainable datasets, and discussing future downside
