Problems occur when we combine Q-learning with a function approximator.

What exactly is the delusional-bias and non-delusional Q-learning? I am talking about the neurIPS 18 best paper Non-delusional Q-learning and value-iteration.

I have trouble understand the term "Policy Commitments", and "consistency". What are they talking about?

PS: a related post

  • 1
    $\begingroup$ Unfortunately I don't think I'll have time to write up a comprehensive answer very soon. A video of the authors' own presentation of their paper at the conference is freely available here though: youtube.com/watch?v=fwSEyyEJmkc Hopefully that can already help you, or anyone else who may want to write up a detailed answer $\endgroup$
    – Dennis Soemers
    Commented Oct 16, 2022 at 10:53
  • $\begingroup$ Non-delusional Q learning is a type of machine learning that can be used to learn how to optimally make decisions in environments where there is some uncertainty. It is related to reinforcement learning, but differs in that it does not require a model of the environment and can be used in environments where the reward function is unknown. $\endgroup$
    – Faizy
    Commented Oct 16, 2022 at 21:10


You must log in to answer this question.