Human Activity Recognition with Unsupervised Learning of Event Logs

Human Activity Recognition (HAR) is a field of study referring to the development of algorithms and systems that can automatically recognize and interpret human activities and associated patterns based on data collected from various sensors. This is particularly crucial for promoting healthier and more independent living, aiding those with disabilities, and providing support to caregivers and medical professionals. In this paper, we propose a framework for analyzing human activities and identifying patterns and disorders with the use of process mining and unsupervised learning, which both have complementary strengths and weaknesses in such problems. Supervised learning is applied on textual data, derived from human activities in smart environments, and incorporates change point detection and clustering algorithms in order to enable the automatic detection of changes in human behavior.