Black-box Safety Analysis and Retraining of DNNs based on Feature Extraction and Clustering

Mohammed Attaoui, Hazem Fahmy, Fabrizio Pastore, Lionel Briand

Research output: Contribution to journalArticlepeer-review

Abstract

Deep neural networks (DNNs) have demonstrated superior performance over classical machine learning to support many features in safety-critical systems. Although DNNs are now widely used in such systems (e.g., self driving cars), there is limited progress regarding automated support for functional safety analysis in DNN-based systems. For example, the identification of root causes of errors, to enable both risk analysis and DNN retraining, remains an open problem. In this article, we propose SAFE, a black-box approach to automatically characterize the root causes of DNN errors. SAFE relies on a transfer learning model pre-trained on ImageNet to extract the features from error-inducing images. It then applies a density-based clustering algorithm to detect arbitrary shaped clusters of images modeling plausible causes of error. Last, clusters are used to effectively retrain and improve the DNN. The black-box nature of SAFE is motivated by our objective not to require changes or even access to the DNN internals to facilitate adoption. Experimental results show the superior ability of SAFE in identifying different root causes of DNN errors based on case studies in the automotive domain. It also yields significant improvements in DNN accuracy after retraining, while saving significant execution time and memory when compared to alternatives.

Original languageEnglish
Article number79
JournalACM Transactions on Software Engineering and Methodology
Volume32
Issue number3
DOIs
Publication statusPublished - 26 Apr 2023
Externally publishedYes

Keywords

  • Additional Key Words and PhrasesDNN explanation
  • DNN debugging
  • DNN functional safety analysis
  • clustering
  • transfer learning

Fingerprint

Dive into the research topics of 'Black-box Safety Analysis and Retraining of DNNs based on Feature Extraction and Clustering'. Together they form a unique fingerprint.

Cite this