Multi Armed Bandits Reinforcement Learning