Concept Enforcement and Modularization for the ISO 26262 Safety Case of Neural Networks

Faculty/Professorship: Cognitive Systems  
Author(s): Schwalbe, Gesina ; Schmid, Ute  
Publisher Information: Bamberg : Otto-Friedrich-Universität
Year of publication: 2020
Pages: 6
Source/Other editions: PhD Forum at the European Conference on Machine Learning and Principles of Knowledge Discovery ECML (2019), 6 S.
Year of first publication: 2019
Language(s): English
DOI: 10.20378/irb-47277
Licence: Creative Commons - CC BY-SA - Attribution - ShareAlike 4.0 International 
URN: urn:nbn:de:bvb:473-irb-472771
The ability to formulate formally verifiable requirements is crucial for the safety verification of software units in the automotive industries. However, it is very restricted for complex perception tasks involving deep neural networks (DNNs) due to their black-box character. For a solution we propose to identify or enforce human interpretable concepts as intermediate output of the DNN. Two effects are expected: Requirements can be formulated using these concepts. And the DNN is modularized, thus reduces complexity and therefore easing a safety case. A research project proposal for a PhD thesis is sketched in the following.
GND Keywords: ISO/DIS 26262; Netzwerk; Verifikation
Keywords: ISO 26262, neural networks, formal verification, concept enforcement
DDC Classification: 004 Computer science  
RVK Classification: ST 300   
Type: Workingpaper
Release Date: 3. July 2020

File Description SizeFormat  
fisba47277.pdf889.15 kBPDFView/Open