Skip to Main Content
The performance of activity recognition systems depends on annotated training data. Obtaining annotated data is a costly and burdensome task. The need for annotated data for activity recognition systems using Hidden Markov models can be reduced by using active learning methods. Active learning lets the learning algorithm to choose the data from which it learns. In this study, uncertainty sampling methods for active learning are shown to reduce the amount of the needed annotated data in an activity recognition task using real data.