Reinforcement Learning and Data-Generation for Syntax-Guided Synthesis

TL;DR

This paper introduces a reinforcement learning approach using Monte-Carlo Tree Search for syntax-guided program synthesis, leveraging automatically generated training data to improve performance and outperform existing tools on benchmark problems.

Contribution

It presents a novel RL-based algorithm for SyGuS that uses MCTS with learned policies and value functions, and introduces a method for generating training data via anti-unification.

Findings

Learned policy improves synthesis performance by over 26 percentage points.

The approach outperforms state-of-the-art tool cvc5 on training data.

Achieves comparable problem-solving rates on benchmarks where cvc5 fails.

Abstract

Program synthesis is the task of automatically generating code based on a specification. In Syntax-Guided Synthesis (SyGuS) this specification is a combination of a syntactic template and a logical formula, and the result is guaranteed to satisfy both. We present a reinforcement-learning guided algorithm for SyGuS which uses Monte-Carlo Tree Search (MCTS) to search the space of candidate solutions. Our algorithm learns policy and value functions which, combined with the upper confidence bound for trees, allow it to balance exploration and exploitation. A common challenge in applying machine learning approaches to syntax-guided synthesis is the scarcity of training data. To address this, we present a method for automatically generating training data for SyGuS based on anti-unification of existing first-order satisfiability problems, which we use to train our MCTS policy. We implement and evaluate this setup and demonstrate that learned policy and value improve the synthesis performance over a baseline by over 26 percentage points in the training and testing sets. Our tool outperforms state-of-the-art tool cvc5 on the training set and performs comparably in terms of the total number of problems solved on the testing set (solving 23% of the benchmarks on which cvc5 fails). We make our data set publicly available, to enable further application of machine learning methods to the SyGuS problem.