Comparative Study of Classifiers on Human Activity Recognition by Different Feature Engineering Techniques

Semi-supervised ensemble learning for human activity recognition in casas Kyoto dataset

“…This could provide a deeper understanding and more accurate classification of human activities. [ 7 ] This paper introduces a wearable sensor-based continuous fall monitoring system designed not only to detect fall incidents but also to identify falling patterns and associated activities, addressing the need for comprehensive fall risk prevention. These proposed models employ a two-layer multi-granularity framework and an emergent paradigm with marker-based stigmergy, enhancing context-aware information aggregation and generating a two-dimensional activity pheromone trail.…”

Human Activity Recognition Data Analysis: History, Evolutions, and New Trends

“…Regarding the application of Machine Learning techniques, to Human Activity Recognition Dataset, various experiments have been developed, but the most relevant ones found in the literature are highlighted below (see Table 3). Tasmin [121], carried out implementations in the UCI-HAR Dataset, through the implementation of supervised algorithms Nearest Neighbor, Decision Tree, Random Forest, and Naive Bayes, of the techniques used, the one with the best results in the detection of activities was the Bayesian with an accuracy of 76.9%. Igwe [122], concentrated his experimentations on the ARAS Data-set which was implemented in 2 different locations (House A and House B), CA-SAS Tulum created by WSU University, the author applied supervised techniques such as SVM, ANN, and MSA (Margin Setting Algorithm), demonstrating the effectiveness of the latter in identifying activities with an accuracy of 68.85%, 96.24% and 68% in the respective Datasets.…”