Publication result detail

Boosted Decision Trees for Behaviour Mining of Concurrent Programs

ŠIMKOVÁ, H.; LETKO, Z.; KŘENA, B.; VOJNAR, T.; DUDKA, V.; AVROS, R.; UR, S.; VOLKOVICH, Z.

Original Title

Boosted Decision Trees for Behaviour Mining of Concurrent Programs

English Title

Boosted Decision Trees for Behaviour Mining of Concurrent Programs

Type

Paper in proceedings outside WoS and Scopus

Original Abstract

Testing of concurrent programs is difficult since the scheduling non-determinism requires one to test a huge number of different thread interleavings. Moreover, a simple repetition of test executions will typically examine similar interleavings only. One popular way how to deal with this
problem is to use the noise injection approach, which is, however, parameterized with many parameters whose suitable values are difficult to find. In this paper,
we propose a novel application of classification-based data mining for this purpose. Our approach can identify which test and noise parameters are the most influential for a given program and a given testing goal and which values (or
ranges of values) of these parameters are suitable for meeting this goal. We present experiments that show that our approach can indeed fully automatically
improve noise-based testing of particular programs with a~particular testing goal. At the same time, we use it to obtain new general insights into noise-based testing as well.

English abstract

Testing of concurrent programs is difficult since the scheduling non-determinism requires one to test a huge number of different thread interleavings. Moreover, a simple repetition of test executions will typically examine similar interleavings only. One popular way how to deal with this
problem is to use the noise injection approach, which is, however, parameterized with many parameters whose suitable values are difficult to find. In this paper,
we propose a novel application of classification-based data mining for this purpose. Our approach can identify which test and noise parameters are the most influential for a given program and a given testing goal and which values (or
ranges of values) of these parameters are suitable for meeting this goal. We present experiments that show that our approach can indeed fully automatically
improve noise-based testing of particular programs with a~particular testing goal. At the same time, we use it to obtain new general insights into noise-based testing as well.

Keywords

Testing, noise injection, classification, AdaBoost, multi-threaded programs

Key words in English

Testing, noise injection, classification, AdaBoost, multi-threaded programs

Authors

ŠIMKOVÁ, H.; LETKO, Z.; KŘENA, B.; VOJNAR, T.; DUDKA, V.; AVROS, R.; UR, S.; VOLKOVICH, Z.

RIV year

2017

Released

31.08.2014

Publisher

NOVPRESS s.r.o.

Location

Brno

ISBN

978-80-214-5022-6

Book

Proceedings of MEMICS'14

Pages from

15

Pages to

27

Pages count

12

BibTex

@inproceedings{BUT111631,
  author="Hana {Šimková} and Zdeněk {Letko} and Bohuslav {Křena} and Tomáš {Vojnar} and Vendula {Dudka} and Renata {Avros} and Shmuel {Ur} and Zeev {Volkovich}",
  title="Boosted Decision Trees for Behaviour Mining of Concurrent Programs",
  booktitle="Proceedings of MEMICS'14",
  year="2014",
  pages="15--27",
  publisher="NOVPRESS s.r.o.",
  address="Brno",
  isbn="978-80-214-5022-6"
}