Understanding and developing intelligent agents that simulate human learning has been a long-standing goal in both artificial intelligence and cognitive science. Although learning agents are able to produce intelligent behavior with less human knowledge engineering than in the past, intelligent agent developers are still required to manually encode much prior domain knowledge. We recently proposed an efficient algorithm that acquires representations of the world using an unsupervised grammar induction algorithm, and integrated this representation learner into a simulated student, SimStudent. In this paper, we use the representation learner to automatically generate a set of feature predicates based on the acquired representation, and provide the automatically generated feature predicates to SimStudent as prior domain knowledge. We show that with the automatically-generated feature predicates, the learning agent can perform at a level comparable to when it is given manually-constructed feature predicates, but without the effort required to create these feature predicates.
展开▼