Adaptive Optimization of Traffic Signal Timing via Deep Reinforcement Learning

Table 3

Simulation environment hyperparameters.

Parameter	Meaning	Value
	Discount factor	0.99
	Learning rate	0.001
	Clip range	0.2
	Every episode simulation time	5000 s
	The number of steps for update	128
	Entropy coefficient for the loss calculator	0.01
	Value function coefficient for the loss function	0.5