Skip to main content

Table 2 Parameters in ACGOP and ACGP

From: Actor-critic learning-based energy optimization for UAV access and backhaul networks

Parameters Actor Critic
Number of hidden layers 3 3
Number of nodes/layer 300 300
Activation function (hidden layers) ReLU ReLU
Activation function (output layer) Sigmoid None
Learning rate \(\rho\) 0.001 0.001
Loss function Eq. (24) Eq. (25)
Optimizer Adam Adam
Batch size 64 64
Discount factor \(\gamma\) 0.9
Size of repository 10,000 tuples
Number of learning episodes 400
Software platform Python 3.6 with
TensorFlow 0.12.1