An algorithm with nearly optimal pseudo-regret for both stochastic and adversarial bandits | Read Paper on Bytez