I want to use reinforcement learning in an environment I made. The exact environment doesn't really matter, but it comes down to this: The amount of different states in the environment is infinite e.g. amount of ways you can put 4 cars at an intersection, but the amount of different actions is only 3 e.g. go forward, right or left. The state exists out of five numbers. My question is: what algorithm should I use or at least what kind of algorithm?
$\begingroup$
$\endgroup$
-
$\begingroup$ Is DQN really good when there are infinite different states the environment can be $\endgroup$ – SirPVP Jul 2 at 14:46
-
$\begingroup$ @SirPVP Yes, it is a function approximator method which is precisely for infinite state spaces. $\endgroup$ – David Ireland Jul 2 at 15:16
-
1
