Achieving complete learning in Multi-Armed Bandit problems | IEEE Conference Publication | IEEE Xplore