Statistical and Computational Trade-off in Multi-Agent Multi-Armed Bandits

Inspired by Mean Field approximation techniques used in graphical models, we provide simple upper bounds of the regret lower bound.