Evolving Structured Programs with Hierarchical Instructions and Skip Nodes

Abstract

To evolve structured programs we introduce H-PIPE, a hierarchical extension of Probabilistic Incremental Program Evolution (PIPE). Structure is induced by "hierarchical instructions" (HIs) limited to top-level, structuring program parts. "Skip nodes" (SNs) allow for switching program parts on and off. They facilitate synthesis of certain structured programs. In our experiments HPIPE outperforms PIPE: structural bias can speed up program synthesis. Keywords: Probabilistic Incremental Program Evolution, Structured Programs, Hierarchical Programs, Non-Coding Segments. 1 Introduction Overview. Automatic program synthesis is of interest because it addresses the problem of searching in general algorithm space as opposed to more limited search spaces like those of, say, feedforward neural networks. Hierarchical Probabilistic Incremental Program Evolution (H-PIPE) is a novel method for synthesizing structured programs. It uses the PIPE paradigm (Sa/lustowicz and Schmidhuber, 1997) to iterativ...