I am trying to train an AI with an environment where the states are continuous but the actions are discrete, that means I can not apply DDPG or TD3.

Can someone please help to let know what should be the best algorithm for discrete action spaces and is there any version of DDPG or TD3 which can be applied to discrete action spaces on partially observable MDPs.

  • $\begingroup$ there are lots of options you can choose from! You might want to start with vanilla DQN and then see if you can use any variants to improve upon this. The same with actor-critic. I usually like to start with the simplest model for my problem and then adapt it until it works. $\endgroup$ Aug 7 '20 at 9:00
  • 1
    $\begingroup$ This resource gives an interactive guide on selection of RL algorithms. intellabs.github.io/coach/selecting_an_algorithm.html It might not give a direct answer for this particular case, but can be useful $\endgroup$
    – mugoh
    Nov 15 '20 at 7:14

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.