Human daily activity recognition with wearable sensors based on incremental learning

Abstract

This paper proposes a human physical activity (PA) recognition method based on incremental learning, which deal with the accuracy loss of traditional recognition system caused by the difference of different individuals. Firstly, the paper introduces the principle of incremental learning, which mainly introduced Learn++ algorithm principle, and describes the differentiation feedback optimization algorithm based on incremental learning in specific details. Then, taking the body sensor network built in the previous work as the data acquisition platform, this article conducts experiments of seven daily activities on five experimental individuals, and verifies the differentiation feedback optimization algorithm of this paper. According to the experiment results, the algorithm described in this paper has an obvious improvement effect on the physical activity recognition performance of the specific individuals.