When dealing with continuous action spaces, a common choice when designing a policy in policy gradient methods is to learn mean and variance of actions for a specific state and then simply sample from the normal distribution defined by the learned mean and variance to get an action.

My first question is, is explicit exploration strategy even needed is such cases, because the dose of randomness in actions would come from the sampling itself, on the other hand there could probably be cases where we would be stuck in a local optimum just by sampling.
My second question is, in case that explicit exploration is needed, how would one approach this problem of exploration for this specific setup.


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.