Implement more best arm identification algorithms

[xuedong]

Have you thought of implementing more BAI algorithms? I can do it, but I'll need some time to get familiar with your environment.

[Naereen]

No, I never though of doing so, but it could be a very valuable addition!!

The entire API is oriented toward the action/reward loop, which is also followed by the BAI task, isn't it?

But the Evaluator class is more oriented towards evaluating the reward maximization task.

[choltz95]

I'm also interested in collaborative best arm identification! Some simple meta-aggregate algorithms like [1, 2] might be a good place to start.
https://arxiv.org/abs/1311.0800
https://arxiv.org/abs/1811.07763

[Naereen]

Hi @choltz95, thanks for your comment!
I am familiar with the second work you quoted, not the first one, thanks for the interesting reference!

I no longer have much time to work on SMPyBandits, and cannot take the time to implement BAI algorithms / benchmarks, unfortunately.
However, I will, of course, do my best to help if you want to take a chance at it!

Regards.