I am currently working on a problem that has 7 continuous actions and instantly gives a reward. I was thinking that there are Contextual-Bandits-Algorithms applicable to this kind of problem, but so far I didn't find any. KR-UCB seemed interesting, but I don't think it is possible to traverse a search tree with multiple continuous actions. I was also thinking about using a DDPG-algorithm, but I heard in terms of regret-optimization those kinds of algorithms aren't too efficient. Maybe there even are some regular non-linear optimizers applicable, which I don't know yet?

  • $\begingroup$ Does an action only affect the reward or also the next state you end up in? If also the next state, then it's not a contextual bandit problem, but the full RL problem. It would be nice if you can clarify this before providing an answer. I think it would be a good idea if you can describe more in detail your specific problem. $\endgroup$
    – nbro
    Dec 21 '21 at 9:48

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.