CORR
2011
Springer
13 years 7 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...