Online activity recognition using evolving classifiers

摘要: This paper describes the use of evolving classifiers for activity recognition from sensor readings in ambient assisted living environments. Recognizing activities an elderly person who lives alone performs, and identifying potential problems detected is a very active topic research. However, current approaches do not take into account fact that way performed by evolves over time therefore are identified mapping them to static model. In this work we describe evaluate approach online classifying based on Evolving Fuzzy Systems (EFS): described model time, according changes observed performed. These have been evaluated three datasets obtained real home settings, achieving good performance, at confidence interval 95%, compared with well know probabilistic models terms F-Measure, but improving their performance capabilities ability adapt ways which carried out.