Table 8 NN model hyperparameters for the adherence prediction experiment

LSTM

Layers: 1 LSTM and 1 Dense

Units: 4 (LSTM) and 2 (Dense)

Batch size: 1024

L2: 9.261e−3

Loss: categorical cross-entropy

Epochs: 100

Simple RNN

Layers: 1 Simple RNN and 1 Dense

Units: 8 (Simple RNN) and 2 (Dense)

Batch size: 4096

L2: 1.202e−2

Loss: categorical cross-entropy

Epochs: 100

MLP

Layers: 3 Dense and 2 Dropout

Units: 64, 64 and 2

Batch size: 256

Dropout rate: Layer 1 2.152e−1

Layer 2 1.758e−1

Loss: categorical cross-entropy

Epochs: 50