Table 3.
The comparison of time consumption among models on the UCI and WISDM datasets.
| Models | Time (ms) | Time (ms) |
|---|---|---|
| on UCI | on WISDM | |
| Inherent Features-based CNN4 | 23.61 | 23.01 |
| Coherent Features-based CNN5 | 25.61 | 24.22 |
| HAR-image CNN7 | 27.46 | 25.31 |
| Multi-stage CNN10 | 54.29 | 49.89 |
| Attention-based multi-head CNN9 | 25.62 | 23.22 |
| Temporal CNN6 | 16.31 | 13.57 |
| Feature Embedding-based LSTM16 | 65.02 | 60.04 |
| ConvAE-LSTM21 | 67.08 | 61.17 |
| Attention-based LSTM22 | 75.18 | 72.61 |
| Hybrid CNN and LSTM17 | 60.03 | 54.14 |
| Proposed Method | 14.71 | 12.11 |