Table 3 Q-network hyperparameter settings
From: Deep reinforcement learning for data-efficient weakly supervised business process anomaly detection
| Â | LSTM | Self-attention | FFNN |
|---|---|---|---|
Input layer size | Trace length * number of activities | Trace length * 512 | Trace length * 512 |
No. of hidden layers | 1 | 0 | 1 |
hidden layer size | 512 | – | Trace length * 1024 |
Output layer size | Trace length * 512 | Trace length * 512 | 2 |
Activation functions | Relu | Softmax | Relu |