Conference item icon

Conference item

Exposing previously undetectable faults in deep neural networks

Abstract:

Existing methods for testing DNNs solve the oracle problem by constraining the raw features (e.g. image pixel values) to be within a small distance of a dataset example for which the desired DNN output is known. But this limits the kinds of faults these approaches are able to detect. In this paper, we introduce a novel DNN testing method that is able to find faults in DNNs that other methods cannot. The crux is that, by leveraging generative machine learning, we can generate fresh test inputs...

Expand abstract
Publication status:
Published
Peer review status:
Peer reviewed

Actions


Access Document


Files:
Publisher copy:
10.1145/3460319

Authors


More by this author
Institution:
University of Oxford
Oxford college:
Balliol College
Role:
Author
ORCID:
0000-0003-1911-7164
More by this author
Institution:
University of Oxford
Division:
MPLS
Department:
Computer Science
Role:
Author
ORCID:
0000-0002-2462-2782
Publisher:
Association for Computing Machinery Publisher's website
Pages:
56–66
Host title:
ISSTA 2021: Proceedings of the 30th ACM SIGSOFT International Symposium on Software Testing and Analysis
Publication date:
2021-07-11
Acceptance date:
2021-04-19
Event title:
ISSTA 2021: ACM SIGSOFT International Symposium on Software Testing and Analysis
Event location:
Virtual Event (Aarhus, Denmark)
Event website:
https://conf.researchr.org/home/issta-2021
Event start date:
2021-07-12T00:00:00Z
Event end date:
2021-07-16T00:00:00Z
DOI:
ISBN:
9781450384599
Language:
English
Keywords:
Pubs id:
1173370
Local pid:
pubs:1173370
Deposit date:
2021-04-25

Terms of use


Views and Downloads






If you are the owner of this record, you can report an update to it here: Report update to this record

TO TOP