Table 1 Properties of the best LSTM model
From: Long short-term memory (LSTM) recurrent neural network for muscle activity detection
LSTM layers | Properties |
|---|---|
Sequence input layer | 1 input feature |
LSTM layer #1 | 275 hidden units Bi-directional sequence-to-sequence architecture |
LSTM layer #2 | 138 hidden units Bi-directional sequence-to-sequence architecture |
Fully connected layer | 2units |
Softmax layer | Softmax activation function (threshold = 0.5) |
Classification output layer | 2 classes (1 = muscle activity, 0 = background noise) |