Hey I am currently taking Stanford cs243 reinforcement learning course in Youtube to learn reinforcement learning in that I understand that policy is something like a function which get a specific state as input and return a action but I am confused that policy will return a action from the state space of the action which likelihood is high now my question is that for deterministic process how the policy chosen and stochastic policy is it return a single action or max(policy(s)=probability distribution of the actions in that specific state) and is that a joint probability or conditional probability?

How to approach a model in RL?

I understood that in a transition model it receives a observation or some information about the world and the state,action if possible now it will give some feedback to agent or a idea how the next state will be or future world idea?

So how to differentiate between a decision model and a reward model?

so help me to validate these to clarify my understanding about the model and policy.


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.