Peter L. Bartlett Varsha Dani Thomas P. Hayes Sham Kakade Alexander Rakhlin Ambuj Tewari High-Probability Regret Bounds for Bandit Online Linear Optimization. 335-342 2008 COLT http://colt2008.cs.helsinki.fi/papers/30-Bartlett.pdf conf/colt/2008 db/conf/colt/colt2008.html#BartlettDHKRT08