COLT
2008
Springer
14 years 10 months ago
2008
Springer
We introduce an efficient algorithm for the problem of online linear optimization in the bandit setting which achieves the optimal O ( T) regret. The setting is a natural general...