DeepCoder: Learning to Write Programs

TL;DR

DeepCoder employs deep learning to predict program properties from input-output examples, significantly accelerating program synthesis and solving competition-level problems more efficiently.

Contribution

This work introduces a neural network-guided search method that enhances program synthesis techniques using deep learning predictions.

Findings

Achieves an order of magnitude speedup over non-augmented baselines

Solves problems of difficulty comparable to simple programming competition tasks

Demonstrates the effectiveness of neural network predictions in guiding search algorithms

Abstract

We develop a first line of attack for solving programming competition-style problems from input-output examples using deep learning. The approach is to train a neural network to predict properties of the program that generated the outputs from the inputs. We use the neural network's predictions to augment search techniques from the programming languages community, including enumerative search and an SMT-based solver. Empirically, we show that our approach leads to an order of magnitude speedup over the strong non-augmented baselines and a Recurrent Neural Network approach, and that we are able to solve problems of difficulty comparable to the simplest problems on programming competition websites.