Shie Mannor Nahum Shimkin Adaptive Strategies and Regret Minimization in Arbitrarily Varying Markov Environments. 128-142 2001 conf/colt/2001 COLT/EuroCOLT http://link.springer.de/link/service/series/0558/bibs/2111/21110128.htm db/conf/colt/colt2001.html#MannorS01