Representing systems with hidden state
Christopher Hundt, Prakash Panagaden, Joëlle Pineau, Doina Precup · 2006
We discuss the problem of finding a good state repre-sentation in stochastic systems with observations. We develop a duality theory that generalizes existing work in predictive state representations as well as automata theory. We discuss how this theoretical framework can be used to build learning algorithms, approximate plan-ning algorithms as well as to deal with continuous ob-servations.