DeepCoder: Learning to Write Programs
arXiv:1611.01989
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.
Submitted to ICLR 2017
Cited by in corpus (18)
- Improved Automatic Summarization of Subroutines via Attention to File Context
- Learning to Represent Programs with Property Signatures
- Sketch2Code: Transformation of Sketches to UI in Real-time Using Deep Neural Network
- Towards Proof Synthesis Guided by Neural Machine Translation for Intuitionistic Propositional Logic
- Synthesizing Database Programs for Schema Refactoring
- Towards Neural-Guided Program Synthesis for Linear Temporal Logic Specifications
- Programmable Agents
- Data Generation for Neural Programming by Example
- Neural Program Synthesis with a Differentiable Fixer
- SynFuzz: Efficient Concolic Execution via Branch Condition Synthesis
- M3: Semantic API Migrations
- Neural Program Synthesis By Self-Learning
- PLANS: Robust Program Learning from Neurally Inferred Specifications
- Using human-in-the-loop synthesis to author functional reactive programs
- Gradient Descent over Metagrammars for Syntax-Guided Synthesis
- Automatically Generating Codes from Graphical Screenshots Based on Deep Autocoder
- Grammar Filtering For Syntax-Guided Synthesis
- Formal Fields: A Framework to Automate Code Generation Across Domains