-
-
Notifications
You must be signed in to change notification settings - Fork 57
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Implement algorithms from "Multiplayer bandits without observing collision information" arXiv:1808.08416 #141
Comments
A few remarks:
|
Yeah, as expected all this work is extremely nice from a theoretical point of view, but quite inpractical for real. Just have a look at the examples of computation of estimated length of uniform exploration phases (1 and 2): >>> estimate_length_phases_12(m=2, K=2, Delta=0.1, T=100)
198214307
>>> estimate_length_phases_12(m=2, K=2, Delta=0.01, T=100)
19821430723
>>> estimate_length_phases_12(m=2, K=2, Delta=0.1, T=1000)
271897030
>>> estimate_length_phases_12(m=2, K=3, Delta=0.1, T=100)
307052623
>>> estimate_length_phases_12(m=2, K=5, Delta=0.1, T=100)
532187397 That's just unreasonable! Too bad 😢 ! |
…ltiplayer bandits without observing collision information', by Gabor Lugosi and Abbas Mehrabian]. Cf. #141
Note that, after discussing with the author, I tried using a much smaller value for g (instead of 128 just 1), and their algorithm is still very much asymptotic. Sadly 😢. But… 😺 we can surely improve it and make it usable in practice, I have no doubt (just no time right now to do it properly). |
The text was updated successfully, but these errors were encountered: