Exploring Classifier Selection for Human Activity Recognition Using Machine Learning Approach

• Published in: 2023 Intelligent Computing and Control for Engineering and Business Systems (ICCEBS) pp. 1 - 5
• Main Authors: Valli, C., Amutha, R.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 14.12.2023
• Subjects: Classification Accuracy; Classification tree analysis; Computational modeling; Confusion Matrix; Human Activity Recognition; KNN; Legged locomotion; Logistic Regression; Older adults; Random Forest; Random forests; Support vector machines; SVM; Wearable sensors
• DOI: 10.1109/ICCEBS58601.2023.10448982

Elderly and differently-abled people usually cherish the ability to stay in their homes and live independently. This independence has its challenges. One of the main threats is falling or having a health disorder at home without anyone's knowledge. Activity recognition plays an important role in supporting the independent living concept. This paper focuses on the best classifier model for activity recognition among various machine learning techniques for the independent living concept, which can be done for the betterment of the lives of elderly and differently abled people. Acquisition of data from the subject is done with tri-axial accelerometer which is present in wearable sensor, undergoes feature extraction to reduce high data rate and redundant nature of information. A well-trained classifier is used to predict one of the basic activities like walking, lying, walking upstairs, walking downstairs, sitting, standing etc. Support Vector Machine (SVM), K- Nearest Neighbor (KNN), Logistic regression (LR) and Random Forest (RF) are the varied Models that are taken for the study and are computed utilizing performance metrics like accuracy, specificity, precision, F-score and sensitivity