Proposing an algorithm for transferring reward samples in sequential multi-armed bandit problems to improve cumulative regret performance.
