Neural-Guided Deductive Search for Real-Time Program Synthesis from Examples

TL;DR

This paper introduces Neural Guided Deductive Search (NGDS), a hybrid program synthesis method combining symbolic logic and neural networks to enable real-time, accurate program generation from few examples, outperforming existing systems.

Contribution

The paper presents NGDS, a novel hybrid approach that integrates deductive logic with neural networks, allowing fast, accurate synthesis from limited data and improving over prior purely symbolic or statistical methods.

Findings

Achieves up to 12x speed-up over state-of-the-art systems

Effectively generalizes to unseen examples

Trains efficiently on limited real-world data

Abstract

Synthesizing user-intended programs from a small number of input-output examples is a challenging problem with several important applications like spreadsheet manipulation, data wrangling and code refactoring. Existing synthesis systems either completely rely on deductive logic techniques that are extensively hand-engineered or on purely statistical models that need massive amounts of data, and in general fail to provide real-time synthesis on challenging benchmarks. In this work, we propose Neural Guided Deductive Search (NGDS), a hybrid synthesis technique that combines the best of both symbolic logic techniques and statistical models. Thus, it produces programs that satisfy the provided specifications by construction and generalize well on unseen examples, similar to data-driven systems. Our technique effectively utilizes the deductive search framework to reduce the learning problem of the neural component to a simple supervised learning setup. Further, this allows us to both train on sparingly available real-world data and still leverage powerful recurrent neural network encoders. We demonstrate the effectiveness of our method by evaluating on real-world customer scenarios by synthesizing accurate programs with up to 12x speed-up compared to state-of-the-art systems.