Improved Regret for Bandit Convex Optimization with Delayed Feedback | Read Paper on Bytez