Publication detail

PAC Learning-Based Verification and Model Synthesis

CHEN, Y. HSIEH, C. LENGÁL, O. LII, T. TSAI, M. WANG, B. WANG, F.

Original Title

PAC Learning-Based Verification and Model Synthesis

Type

conference paper

Language

English

Original Abstract

We introduce a novel technique for verification and model synthesis of sequential programs. Our technique is based on learning an approximate regular model of the set of feasible paths in a program, and testing whether this model contains an incorrect behavior. Exact learning algorithms require checking equivalence between the model and the program, which is a difficult problem, in general undecidable. Our learning procedure is therefore based on the framework of probably approximately correct (PAC) learning, which uses sampling instead, and provides correctness guarantees expressed using the terms error probability and confidence. Besides the verification result, our procedure also outputs the model with the said correctness guarantees. Obtained preliminary experiments show encouraging results, in some cases even outperforming mature software verifiers.

Keywords

model synthesis, PAC learning, finite automata, program verification

Authors

CHEN, Y.; HSIEH, C.; LENGÁL, O.; LII, T.; TSAI, M.; WANG, B.; WANG, F.

Released

16. 2. 2016

Publisher

Association for Computing Machinery

Location

Austin, TX

ISBN

978-1-4503-3900-1

Book

Proceedings of the 38th International Conference on Software Engineering

Pages from

714

Pages to

724

Pages count

11

URL

http://dx.doi.org/10.1145/2884781.2884860

BibTex

@inproceedings{BUT130941,
  author="Yu-Fang {Chen} and Chiao {Hsieh} and Ondřej {Lengál} and Tsung-Ju {Lii} and Ming-Hsien {Tsai} and Bow-Yaw {Wang} and Farn {Wang}",
  title="PAC Learning-Based Verification and Model Synthesis",
  booktitle="Proceedings of the 38th International Conference on Software Engineering",
  year="2016",
  pages="714--724",
  publisher="Association for Computing Machinery",
  address="Austin, TX",
  doi="10.1145/2884781.2884860",
  isbn="978-1-4503-3900-1",
  url="http://dx.doi.org/10.1145/2884781.2884860"
}