Deterministic Bellman residual minimization is the problem of learning values in a Markov Decision Process (MDP) by optimizing for the Bellman residuals directly without using any heuristic surrogates. Compared to the ...