KIT | KIT-Bibliothek | Impressum | Datenschutz

EFFECT: An End-to-End Framework for Evaluating Strategies for Parallel AI Anomaly Detection

Stammler, Matthias 1; Höfer, Julian ORCID iD icon 1; Kraus, David ORCID iD icon 2; Schmidt, Patrick 1; Hotfilter, Tim ORCID iD icon 1; Harbaum, Tanja ORCID iD icon 1; Becker, Jürgen 1
1 Institut für Technik der Informationsverarbeitung (ITIV), Karlsruher Institut für Technologie (KIT)
2 Karlsruher Institut für Technologie (KIT)

Abstract:

Neural networks achieve high accuracy in tasks like image recognition or segmentation. However, their application in safety-critical domains is limited due to their black-box nature and vulnerability to specific types of attacks. To mitigate this, methods detecting out-of-distribution or adversarial attacks in parallel to the network inference were introduced. These methods are hard to compare because they were developed for different use cases, datasets, and networks. To fill this gap, we introduce EFFECT, an end-to-end framework to evaluate and compare new methods for anomaly detection, without the need for retraining and by using traces of intermediate inference results. The presented workflow works with every preexisting neural network architecture and evaluates the considered anomaly detection methods in terms of accuracy and computational complexity. We demonstrate EFFECT's capabilities, by creating new detectors for ShuffleNet and MobileNetV2 for anomaly detection as well as fault origin detection. EFFECT allows us to design an anomaly detector, based on the Mahalanobis distance as well as CNN based detectors. For both use cases, we achieve accuracies of over 85 %, classifying inferences as normal or abnormal, and thus beating existing methods.


Verlagsausgabe §
DOI: 10.5445/IR/1000164440
Veröffentlicht am 21.11.2023
Originalveröffentlichung
DOI: 10.1016/j.procs.2023.08.188
Scopus
Zitationen: 1
Dimensions
Zitationen: 2
Cover der Publikation
Zugehörige Institution(en) am KIT Institut für Technik der Informationsverarbeitung (ITIV)
Publikationstyp Zeitschriftenaufsatz
Publikationsjahr 2023
Sprache Englisch
Identifikator ISSN: 1877-0509
KITopen-ID: 1000164440
Erschienen in Procedia Computer Science
Verlag Elsevier
Band 222
Seiten 499 – 508
Bemerkung zur Veröffentlichung Part of special issue: International Neural Network Society Workshop on Deep Learning Innovations and Applications (INNS DLIA 2023)
Vorab online veröffentlicht am 31.08.2023
Schlagwörter AI security, inference monitoring, layer tracing, evaluation framework, anomaly detection
Nachgewiesen in Dimensions
Scopus
KIT – Die Forschungsuniversität in der Helmholtz-Gemeinschaft
KITopen Landing Page