Machine learning methods for human activity recognition using ambient sensors

Human activity recognition is a fast growing research field with the goal to identify human behavior based on collected data. Human activity recognition systems have many applications in healthcare, sport, manufacturing and other spheres. Data for activity recognition may be collected from video-cameras or various sensors. Ambient sensors have a lot of advantages such as simplicity, inexpensiveness and wide application in smart-houses. Machine learning methods are often used for pattern activity recognition in sensor data. This work presents an application and comparison of some classical, ensemble and neural network machine learning methods for human activity recognition using smart-house ambient sensor data.