Type-driven Neural Programming by Example

Programming by example (PBE) has traditionally seen a split between formal versus neural approaches, meaning programming types had yet to be used in neural program synthesis. We propose a way to incorporate programming types into a neural program synthesis approach for programming by example. We propose the Typed Neuro-Symbolic Program Synthesis (TNSPS) method based on this idea, and test it in the functional programming context to empirically verify type information may help improve generalization in neural synthesizers on limited-size datasets. Finally we discuss several topics of interest that may help take these ideas further. For reproducibility, we release our code publicly.