Gradient-based Reinforcement Planning in Policy-Search Methods

What this record says

We introduce a learning method called ``gradient-based reinforcement planning'' (GREP). Unlike traditional DP methods that improve their policy backwards in

CATALOG · The abstract is indexed.