Black-box Safety Analysis and Retraining of DNNs based on Feature Extraction and Clustering
This addresses the need for automated functional safety analysis in DNN-based systems like self-driving cars, offering a practical solution for error diagnosis and retraining without requiring access to DNN internals.
The paper tackles the problem of identifying root causes of errors in deep neural networks (DNNs) for safety-critical systems, proposing SAFE, a black-box approach that uses feature extraction and clustering to characterize errors and retrain DNNs, resulting in significant improvements in accuracy and efficiency.
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 paper, 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.