A Deep Learning-Based Framework for Human Activity Recognition in Smart Homes

Table 4

Results of all models in the “all-features approach.”

All-features approach (258 features)	Accuracy (%)	F-measure (%)	Training time (sec)	Testing time (sec)
Baseline-RF	92.22	92.83	12.21	0.96
MLP	94.26	94.73	61.99	3.13
RNN	94.20	94.74	84.83	3.71
LSTM	93.53	94.27	153.06	5.48
GRU	93.40	94.31	153.94	5.13