Skip to main content
. 2018 Oct 25;18(11):3629. doi: 10.3390/s18113629

Figure 5.

Figure 5

The proposed hierarchical activity recognition framework with feature selection integrated. In the phase of training, it selects the optimal feature subset and builds a classifier for each of the non-leaf nodes. In the phase of prediction, it adopts a top-down fashion to give the predicted label.