Table 6 Classification accuracies attained for each HAR datasets before and after applying BBA based feature subset selection.
Dataset | Model | Accuracy without wrapper(%) | Accuracy after wrapper(%) |
|---|---|---|---|
HARTH | Mobilenet_v3_large | 60 | 87 |
Efficientnet_b0 | 68 | 86 | |
Concatenated feature vector | 68 | 88.89 | |
KU-HAR | Mobilenet_v3_large | 73 | 97 |
Efficientnet_b0 | 78 | 96 | |
Concatenated feature vector | 77 | 97.97 | |
HuGaDB | Mobilenet_v3_large | 88 | 93 |
Efficientnet_b0 | 88 | 92 | |
Concatenated feature vector | 88 | 93.82 |
