Specifying Goals to Deep Neural Networks with Answer Set Programming
Abstract: Methods such as DeepCubeA have used deep reinforcement learning to learn domain-specific heuristic functions in a largely domain-independent fashion to solve planning problems. However,…