Some illustrations for this project

Here are some plots illustrating the performances of the different policies implemented in this project, against various problems (with Bernoulli arms only):

(Average) cumulative regret

100000_steps__50_repetition_6_policies_1_Aggr.png

(Average) cumulative regret and standard deviation - FIXME

100000_steps__50_repetition_6_policies_1_Aggr__mean_and_std.png

Normalized cumulative regret

100000_steps__50_repetition_6_policies_1_Aggr__normalized.png

Best arm pulls frequency

100000_steps__50_repetition_6_policies_1_Aggr__bestArmFrequency.png