Cross-mobile ELM based Activity Recognition

Activity recognition using mobile phones has great potential in many applications including mobile healthcare. In order to let a person easily know whether he is in strict compliance with the doctor's exercise prescription and adjust his exercise amount accordingly, we can use a mobile phone based activity reporting system to accurately recognize a range of daily activities and report the duration of each activity. A triaxial accelerometer built-in the mobile phone is used for the classification of several activities, such as staying still, walking, running, and going upstairs and downstairs. To build an activity recognition model, we usually employ one or some specific persons and a specific mobile phone to collect the training samples. However, the world doesn't have the same two mobile phones. The model learnt on one mobile phone may perform poor on another one due to the different offset o, sensitivity s and sampling frequency f values. To solve the cross-mobile problem, we propose an algorithm known as TransELMAR(Transfer learning and Extreme Learning Machine based Activity Recognition) that integrates the transfer learning technique and extreme learning machine algorithm for activity recognition model adaptation. Tested on a real-world data set, the results show that our algorithm outperforms several traditional baseline algorithm.