Table 1 Table of hyperparameters.

From: A Novel CNN-based Bi-LSTM parallel model with attention mechanism for human activity recognition with noisy data

Hyper-parameters

Values

Number of parallel CNN-based Bi-LSTM units

80% of CPU cores

Hidden neurons of 1D CNN units

32

Batch size

16

Dropout ratio

0.6

Dominant layers

Bi-LSTM, Attention, Dropout

Back to article page