4 ms·
Challenges of real-world reinforcement learning
- orasis 7y agoReinforcement learning without simulation is not feasible for complex problems. If you can’t simulate, stick to multi-armed bandits.
- throwlaplace 7y agoanyone know a good survey paper about current SOA? i've read sutton&barton, skimmed openai "Spinning Up in Deep RL". anything else out there?
- hcrisp 7y agoThe Deep RL Bootcamp lectures are still good and relevant: https://sites.google.com/view/deep-rl-bootcamp/lectures https://sites.google.com/view/deep-rl-bootcamp/lectures
- throwlaplace 7y agobeggars can't be choosers of course but i would prefer something to read rather than something to watch
- orasis 7y agohttps://spinningup.openai.com/en/latest/spinningup/keypapers.html https://spinningup.openai.com/en/latest/spinningup/keypapers...