Learning from Delayed Semi-Bandit Feedback under Strong Fairness Guarantees | IEEE Conference Publication | IEEE Xplore