Preferential Temporal Difference Learning

                      Nishanth Anand 1 2 Doina Precup 1 2 3

             Abstract                 TD-learning can be viewed as a way to approximate dy-
    Temporal-Difference (TD) learning is a general       namic programming algorithms in Markovian environ-
    and very useful tool for estimating the value func-     ments (Barnard, 1993). But, if the Markovian assumption
    tion of a given policy, which in turn is required      does not hold (as is  ...