SpecRepair : Counter-Example Guided Safety Repair of Deep Neural Networks

Bauer-Marquart, Fabian; Boetius, David; Leue, Stefan; Schilling, Christian

SpecRepair : Counter-Example Guided Safety Repair of Deep Neural Networks

Dateien

Bauer-Marquart_2-19tqdmlko0rgt3.PDFGröße: 441.19 KBDownloads: 14

Datum

2022

Publikationstyp

Beitrag zu einem Konferenzband

Publikationsstatus

Published

Erschienen in

LEGUNSEN, Owolabi, ed., Grigore ROSU, ed.. Model Checking Software : 28th International Symposium, SPIN 2022, Virtual Event, May 21, 2022 : Proceedings. Cham: Springer, 2022, pp. 79-96. ISSN 0302-9743. eISSN 1611-3349. ISBN 978-3-031-15076-0. Available under: doi: 10.1007/978-3-031-15077-7_5

Zusammenfassung

Deep neural networks (DNNs) are increasingly applied in safety-critical domains, such as self-driving cars, unmanned aircraft, and medical diagnosis. It is of fundamental importance to certify the safety of these DNNs, i.e. that they comply with a formal safety specification. While safety certification tools exactly answer this question, they are of no help in debugging unsafe DNNs, requiring the developer to iteratively verify and modify the DNN until safety is eventually achieved. Hence, a repair technique needs to be developed that can produce a safe DNN automatically. To address this need, we present SpecRepair, a tool that efficiently eliminates counter-examples from a DNN and produces a provably safe DNN without harming its classification accuracy. SpecRepair combines specification-based counter-example search and resumes training of the DNN, penalizing counter-examples and certifying the resulting DNN. We evaluate SpecRepair’s effectiveness on the ACAS Xu benchmark, a DNN-based controller for unmanned aircraft, and two image classification benchmarks. The results show that SpecRepair is more successful in producing safe DNNs than comparable methods, has a shorter runtime, and produces safe DNNs while preserving their classification accuracy.

Fachgebiet (DDC)

004 Informatik

Schlagwörter

Neural networks, Safety repair, Safety specification

Konferenz

Model Checking Software : 28th International Symposium, SPIN 2022 (virtual), 21. Mai 2022

Zitieren

ISO 690

BAUER-MARQUART, Fabian, David BOETIUS, Stefan LEUE, Christian SCHILLING, 2022. SpecRepair : Counter-Example Guided Safety Repair of Deep Neural Networks. Model Checking Software : 28th International Symposium, SPIN 2022 (virtual), 21. Mai 2022. In: LEGUNSEN, Owolabi, ed., Grigore ROSU, ed.. Model Checking Software : 28th International Symposium, SPIN 2022, Virtual Event, May 21, 2022 : Proceedings. Cham: Springer, 2022, pp. 79-96. ISSN 0302-9743. eISSN 1611-3349. ISBN 978-3-031-15076-0. Available under: doi: 10.1007/978-3-031-15077-7_5

BibTex

@inproceedings{BauerMarquart2022SpecR-59172,
  year={2022},
  doi={10.1007/978-3-031-15077-7_5},
  title={SpecRepair : Counter-Example Guided Safety Repair of Deep Neural Networks},
  number={13255},
  isbn={978-3-031-15076-0},
  issn={0302-9743},
  publisher={Springer},
  address={Cham},
  series={Lecture Notes in Computer Science},
  booktitle={Model Checking Software : 28th International Symposium, SPIN 2022, Virtual Event, May 21, 2022 : Proceedings},
  pages={79--96},
  editor={Legunsen, Owolabi and Rosu, Grigore},
  author={Bauer-Marquart, Fabian and Boetius, David and Leue, Stefan and Schilling, Christian}
}

RDF

<rdf:RDF
    xmlns:dcterms="http://purl.org/dc/terms/"
    xmlns:dc="http://purl.org/dc/elements/1.1/"
    xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
    xmlns:bibo="http://purl.org/ontology/bibo/"
    xmlns:dspace="http://digital-repositories.org/ontologies/dspace/0.1.0#"
    xmlns:foaf="http://xmlns.com/foaf/0.1/"
    xmlns:void="http://rdfs.org/ns/void#"
    xmlns:xsd="http://www.w3.org/2001/XMLSchema#" > 
  <rdf:Description rdf:about="https://kops.uni-konstanz.de/server/rdf/resource/123456789/59172">
    <dc:rights>terms-of-use</dc:rights>
    <dcterms:available rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2022-11-16T12:02:29Z</dcterms:available>
    <dc:language>eng</dc:language>
    <dcterms:hasPart rdf:resource="https://kops.uni-konstanz.de/bitstream/123456789/59172/1/Bauer-Marquart_2-19tqdmlko0rgt3.PDF"/>
    <dcterms:abstract xml:lang="eng">Deep neural networks (DNNs) are increasingly applied in safety-critical domains, such as self-driving cars, unmanned aircraft, and medical diagnosis. It is of fundamental importance to certify the safety of these DNNs, i.e. that they comply with a formal safety specification. While safety certification tools exactly answer this question, they are of no help in debugging unsafe DNNs, requiring the developer to iteratively verify and modify the DNN until safety is eventually achieved. Hence, a repair technique needs to be developed that can produce a safe DNN automatically. To address this need, we present SpecRepair, a tool that efficiently eliminates counter-examples from a DNN and produces a provably safe DNN without harming its classification accuracy. SpecRepair combines specification-based counter-example search and resumes training of the DNN, penalizing counter-examples and certifying the resulting DNN. We evaluate SpecRepair’s effectiveness on the ACAS Xu benchmark, a DNN-based controller for unmanned aircraft, and two image classification benchmarks. The results show that SpecRepair is more successful in producing safe DNNs than comparable methods, has a shorter runtime, and produces safe DNNs while preserving their classification accuracy.</dcterms:abstract>
    <dcterms:title>SpecRepair : Counter-Example Guided Safety Repair of Deep Neural Networks</dcterms:title>
    <dc:contributor>Schilling, Christian</dc:contributor>
    <bibo:uri rdf:resource="https://kops.uni-konstanz.de/handle/123456789/59172"/>
    <dc:contributor>Leue, Stefan</dc:contributor>
    <dc:creator>Schilling, Christian</dc:creator>
    <dcterms:issued>2022</dcterms:issued>
    <dc:creator>Bauer-Marquart, Fabian</dc:creator>
    <dcterms:isPartOf rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
    <foaf:homepage rdf:resource="http://localhost:8080/"/>
    <dspace:isPartOfCollection rdf:resource="https://kops.uni-konstanz.de/server/rdf/resource/123456789/36"/>
    <void:sparqlEndpoint rdf:resource="http://localhost/fuseki/dspace/sparql"/>
    <dc:contributor>Boetius, David</dc:contributor>
    <dcterms:rights rdf:resource="https://rightsstatements.org/page/InC/1.0/"/>
    <dc:date rdf:datatype="http://www.w3.org/2001/XMLSchema#dateTime">2022-11-16T12:02:29Z</dc:date>
    <dc:contributor>Bauer-Marquart, Fabian</dc:contributor>
    <dspace:hasBitstream rdf:resource="https://kops.uni-konstanz.de/bitstream/123456789/59172/1/Bauer-Marquart_2-19tqdmlko0rgt3.PDF"/>
    <dc:creator>Boetius, David</dc:creator>
    <dc:creator>Leue, Stefan</dc:creator>
  </rdf:Description>
</rdf:RDF>

Universitätsbibliographie

Ja