Top program construction and reduction for polynomial time Meta-Interpretive learning

Meta-Interpretive Learners, like most ILP systems, learn by searching for a correct hypothesis in the space, powerset of all constructible clauses. We show how this exponentially-growing search can be replaced construction Top program: set clauses hypotheses that is itself hypothesis. give an algorithm program and it constructs polynomial time from finite number examples. implement our Prolog as basis new MIL system, Louise, then reduces removing redundant compare Louise to state-of-the-art search-based system Metagol experiments on grid world navigation, graph connectedness grammar learning datasets find improves Metagol's predictive accuracy when space target theory are both large, or does not include because "classification noise" form mislabelled When small, perform equally well.