Generating correctness proofs with neural networks
Generating correctness proofs with neural networks
复制标题
使用神经网络生成正确性证明
DOI:
10.1145/3394450.3397466
复制
发表时间:
2020
期刊:
影响因子:
--
通讯作者:
Lerner, Sorin
中科院分区:
文献类型:
--
作者:
Sanchez-Stern, Alex;Alhessi, Yousef;Saul, Lawrence;Lerner, Sorin
Foundational verification allows programmers to build software which has been empirically shown to have high levels of assurance in a variety of important domains. However, the cost of producing foundationally verified software remains prohibitively high for most projects, as it requires significant manual effort by highly trained experts. In this paper we present Proverbot9001, a proof search system using machine learning techniques to produce proofs of software correctness in interactive theorem provers. We demonstrate Proverbot9001 on the proof obligations from a large practical proof project, the CompCert verified C compiler, and show that it can effectively automate what were previously manual proofs, automatically producing proofs for 28% of theorem statements in our test dataset, when combined with solver-based tooling. Without any additional solvers, we exhibit a proof completion rate that is a 4X improvement over prior state-of-the-art machine learning models for generating proofs in Coq.
登录
查看更多内容
影响因子:
--
作者:
Heras J
通讯作者:
Heras J
DOI:
--
发表时间:
2019
期刊:
International Conference on Machine Learning
影响因子:
--
作者:
Yang, Kaiyu;Deng, Jia
通讯作者:
Deng, Jia
DOI:
--
发表时间:
2018
期刊:
影响因子:
--
作者:
Daniel Kästner;J. Barrho;Ulrich Wünsche;Marc Schlickling;Bernhard Schommer;Michael Schmidt;C. Ferdinand;X. Leroy;Sandrine Blazy
通讯作者:
Sandrine Blazy
DOI:
10.1007/s10817-018-9458-4
发表时间:
2018
期刊:
Journal of automated reasoning
影响因子:
--
作者:
Czajka Ł;Kaliszyk C
通讯作者:
Kaliszyk C
影响因子:
3.4
作者:
Ekaterina Komendantskaya;Jónathan Heras;G. Grov
通讯作者:
Ekaterina Komendantskaya;Jónathan Heras;G. Grov