Learning to select examples for program synthesis

Program synthesis is a class of regression problems where one seeks a solution, in the form of a source-code program, that maps the inputs to their corresponding outputs exactly. Due to its precise and combinatorial nature, it is commonly formulated as a constraint satisfaction problem, where input-output examples are expressed constraints, and solved with a constraint solver. A key challenge of this formulation is that of scalability: While constraint solvers work well with few well-chosen examples, constraining the entire set of example constitutes a significant overhead in both time and memory. In this paper we address this challenge by constructing a representative subset of examples that is both small and is able to constrain the solver sufficiently. We build the subset one example at a time, using a trained discriminator to predict the probability of unchosen input-output examples conditioned on the chosen input-output examples, adding the least probable example to the subset. Experiment on a diagram drawing domain shows our approach produces subset of examples that are small and representative for the constraint solver.