Code2Inv: A Deep Learning Framework for Program Verification
Authors:
Open Access
- 14 July 2020
- book chapter
- Published by Springer Nature
- Vol. 12225, 151-164
Abstract
We propose a general end-to-end deep learning framework Code2Inv, which takes a verification task and a proof checker as input, and automatically learns a valid proof for the verification task by interacting with the given checker. Code2Inv is parameterized with an embedding module and a grammar: the former encodes the verification task into numeric vectors while the latter describes the format of solutions Code2Inv should produce. We demonstrate the flexibility of Code2Inv by means of two small-scale yet expressive instances: a loop invariant synthesizer for C programs, and a Constrained Horn Clause (CHC) solver.Keywords
This publication has 24 references indexed in Scilit:
- Continuously reasoning about programs using differential Bayesian inference2019
- Search-based program synthesisCommunications of the ACM, 2018
- Scaling Enumerative Program Synthesis via Divide and Conquer2017
- Learning invariants using decision trees and implication counterexamples2016
- Horn Clause Solvers for Program Verification2015
- The SeaHorn Verification Framework2015
- ICE: A Robust Framework for Learning Invariants2014
- Syntax-guided synthesis2013
- Oracle-guided component-based program synthesis2010
- Long Short-Term MemoryNeural Computation, 1997