5 ms·
https://www.alexirpan.com/2018/02/14/rl-hard.html https://www.alexirpan.com/2018/02/14/rl-hard.html Reinforcment learning for the average person is a big waste
by crapflare 8y ago
https://www.alexirpan.com/2018/02/14/rl-hard.html https://www.alexirpan.com/2018/02/14/rl-hard.html Reinforcment learning for the average person is a big waste of time. Probably for anyone atm
- bitL 8y agoDRL is fun, that's what matters! :) Though classical non-biological RL has some strong assumptions (Markovian) that may not work in real world (it's nice to see fixed points with Bellman operator in theory), but for some reason with the "Deep" part, magic happens.
- kevinwang 8y agoI disagree with the conclusion. Article has some critiques on DRL, but I don't think those invalidate the field as a whole. Even the article has a section for "When could Deep RL work for me?"
- alexgmcm 8y agoAlso where do you draw the line? I mean outside of DRL you still have multi-armed bandits which perform very well in industry and RL also inherits a lot from optimal control theory which is used in robotics etc. It isn't that surprising that the cutting edge of the field still has problems - if it didn't then the field would be dead.