Fraunhofer-Gesellschaft

Publica

Hier finden Sie wissenschaftliche Publikationen aus den Fraunhofer-Instituten.

Inspect, Understand, Overcome: A Survey of Practical Methods for AI Safety

 
: Houben, Sebastian; Abrecht, Stephanie; Akila, Maram; Bär, Andreas; Brockherde, Felix; Feifel, Patrick; Fingscheidt, Tim; Gannamaneni, Sujan Sai; Ghobadi, Seyed Eghbal; Hammam, Ahmed; Haselhoff, Anselm; Hauser, Felix; Heinzemann, Christian; Hoffmann, Marco; Kapoor, Nikhil; Kappel, Falk; Klingner, Marvin; Kronenberger, Jan; Küppers, Fabian; Löhdefink, Jonas; Mlynarski, Michael; Mock, Michael; Mualla, Firas; Pavlitskaya, Svetlana; Poretschkin, Maximilian; Pohl, Alexander; Ravi-Kumar, Varun; Rosenzweig, Julia; Rottmann, Matthias; Rüping, Stefan; Sämann, Timo; Schneider, Jan David; Schulz, Elena; Schwalbe, Gesina; Sicking, Joachim; Srivastava, Toshika; Varghese, Serin; Weber, Michael; Wirkert, Sebastian; Wirtz, Tim; Woehrle, Matthias

:
Volltext urn:nbn:de:0011-n-6346547 (645 KByte PDF)
MD5 Fingerprint: ba5818c4c37b2f9da43e464d7ce7f106
Erstellt am: 6.5.2021


Online im WWW, 2021, arXiv:2104.14235, 94 S.
Bundesministerium fur Wirtschaft und Energie BMWi (Deutschland)
VDA Leitinitiative autonomes und vernetztes Fahren; 19A19005X
KI Absicherung - Safe AI for Automated Driving
Bundesministerium für Bildung und Forschung BMBF (Deutschland)
01IS18038B; ML2R
Machine-Learning Rhein-RuhR
Englisch
Bericht, Elektronische Publikation
Fraunhofer IAIS ()
trustworthy AI; machine learning; automated driving; Safety Concerns; Safe AI; AI Safeguarding; KI Absicherung

Abstract
The use of deep neural networks (DNNs) in safety-critical applications like mobile health and autonomous driving is challenging due to numerous model-inherent shortcomings. These shortcomings are diverse and range from a lack of generalization over insufficient interpretability to problems with malicious inputs. Cyber-physical systems employing DNNs are therefore likely to suffer from safety concerns. In recent years, a zoo of state-of-the-art techniques aiming to address these safety concerns has emerged. This work provides a structured and broad overview of them. We first identify categories of insufficiencies to then describe research activities aiming at their detection, quantification, or mitigation. Our paper addresses both machine learning experts and safety engineers: The former ones might profit from the broad range of machine learning (ML) topics covered and discussions on limitations of recent methods. The latter ones might gain insights into the specifics of modern ML methods. We moreover hope that our contribution fuels discussions on desiderata for ML systems and strategies on how to propel existing approaches accordingly.

: http://publica.fraunhofer.de/dokumente/N-634654.html