134
click to vote
CORR
2011
Springer
14 years 10 months ago
2011
Springer
We consider finite horizon Markov decision processes under performance measures that involve both the mean and the variance of the cumulative reward. We show that either randomiz...