Write a Blog >>
ICST 2022
Mon 4 - Fri 8 April 2022
Wed 6 Apr 2022 16:00 - 16:15 at Margaret Hamilton - ICST AI II Chair(s): Donghwan Shin

Software verification is the task of proving correctness of programs against specified requirements. Key to software verification is the automatic generation of loop invariants. In recent years, template- and logic-based approaches to invariant generation have been complemented by machine learning (ML) techniques. A number of proposals for such techniques exist today. Although all authors perform experimental evaluations of their proposals, comparability of the core techniques is nevertheless hindered by differing benchmarks, specific tunings of hyperparameters, missing public availability as well as specialized preprocessings and runtime environments.

In this paper, we present the modular framework MIGML for experimentation with and comparison of ML invariant generators. MIGML contains the core ingredients of ML based invariant generators (i.e. a teacher and a learner) as instantiable components with clear-cut interfaces. This conceptually novel framework allows for a reproducibility study of four existing ML invariant generators: we re-implement the teacher and learner components of the four techniques within our framework which permits a comparison on equal grounds. We are able to successfully reproduce and partially confirm the reported results. We furthermore experiment with novel combinations of components, e.g. employ the data generator within the teacher of technique A together with the learner of technique B. As a result, we observe that such combinations can lead to an overall enhanced effectiveness.

Wed 6 Apr

Displayed time zone: Amsterdam, Berlin, Bern, Rome, Stockholm, Vienna change

15:15 - 16:30
ICST AI IITesting Tools / Research Papers / Industry at Margaret Hamilton
Chair(s): Donghwan Shin University of Luxembourg
15:15
15m
Talk
Learning Realistic Mutations: Bug Creation for Neural Bug Detectors
Research Papers
Cedric Richter Carl von Ossietzky Universität Oldenburg / University of Oldenburg, Heike Wehrheim Carl von Ossietzky Universität Oldenburg / University of Oldenburg
15:30
15m
Talk
SUPERNOVA: Automating Test Selection and Defect Prevention in AAA Video Games Using Risk Based Testing and Machine Learning
Industry
Alexander Senchenko Electronics Arts, Naomi Patterson Electronics Arts, Hamman Samuel Electronics Arts, Dan Ispir Electronics Arts
Pre-print
15:45
15m
Talk
RiverGame - a game testing tool using artificial intelligence
Testing Tools
Ciprian Paduraru University of Bucharest, Miruna Gabriela Paduraru University of Bucharest , Alin Stefanescu University of Bucharest
16:00
15m
Talk
Machine Learning Based Invariant Generation: A Framework and Reproducibility Study
Research Papers
Jan Haltermann University of Oldenburg, Heike Wehrheim Carl von Ossietzky Universität Oldenburg / University of Oldenburg
16:15
15m
Live Q&A
Discussion and Q&A
Research Papers