3 ms·
As someone who studies reinforcement learning (or at least attempts to), this article was expertly written and illustrated. The graphics were phenomenal to expl
by hervature 9y ago
As someone who studies reinforcement learning (or at least attempts to), this article was expertly written and illustrated. The graphics were phenomenal to explain how MDP's function. The text was written with enough intuitive explanations to guide the reader but provides the correct notation to the reader to further investigate. If I had any complaint, it would be that the author didn't mention backwards induction or dynamic programming when describing the solution method in Appendix B. Still, I think this was the best intro/case study of standalone MDP's I have ever read. Great work!
- jdleesmiller 9y ago(Author here.) Thanks for the kind words! I'm glad to hear that. I hope this article will do some good by introducing more people to MDPs --- they are incredibly useful. That is a good point about mentioning backwards induction in Appendix B, because that's exactly what I used! I'll add a note about it.
- Cyphase 9y agoI haven't read the article yet, but I'm definitely going to; what a great review! This part is the kicker: > The text was written with enough intuitive explanations to guide the reader but provides the correct notation to the reader to further investigate. So many explanations about complex or in-depth topics have difficulty finding the sweet spot between being too basic and hand-wavy on one end of the spectrum, and too obtuse and difficult to follow for a layperson on the other end. I've particularly noticed this with articles on quantum mechanics and relativity in the past. It's either complex mathematical equations or trains and flashlights (I'm exaggerating a bit). When I read an article on such a topic that is accurate, yet understandable, while giving the reader the information and keywords they need in order to learn more, and providing a good on-ramp into the topic, it's a noticeable experience.