Research Article
Walking Gait Phase Detection Based on Acceleration Signals Using Voting-Weighted Integrated Neural Network
Table 2
Parameter setting of VWI-DNN structure.
| Layer | Number of neurons | Activation function | Dropout | SNN_1 | SNN_2 | SNN_3 | SNN_1 | SNN_2 | SNN_3 | SNN_1 | SNN_2 | SNN_3 |
| HL_1 | 540 | 540 | 540 | leaky_relu | leaky_relu | leaky_relu | 1.0 | 1.0 | 1.0 | HL_2 | 200 | 200 | 200 | leaky_relu | leaky_relu | leaky_relu | 1.0 | 1.0 | 1.0 | HL_3 | 260 | 200 | 140 | leaky_relu | relu | tanh | 0.6 | 0.8 | 1.0 | HL_4 | 120 | 90 | 60 | leaky_relu | relu | tanh | 1.0 | 1.0 | 1.0 | HL_5 | 60 | 90 | — | leaky_relu | relu | — | 1.0 | 1.0 | — |
|
|