2021/04/22 by Julia Rosenzweig, Rosenzweig, Julia, Joachim Sicking +7
Computer Science · Medicine · #Adversarial Robustness in Machine Learning #Artificial Intelligence in Healthcare and Education #Computer Vision and Pattern Recognition (cs.CV) #Cryptography and Security (cs.CR) #Explainable Artificial Intelligence (XAI) #FOS: Computer and information sciences #Machine Learning (cs.LG) #cs.CR #cs.CV #cs.LG
paper · pdf · doi:10.48550/arxiv.2104.11691
Under IEEE Copyright; accepted at the SAIAD (Safe Artificial Intelligence for Automated Driving) Workshop at CVPR 2021
arxiv created 2021/04/22 · openalex publication_date 2021/04/22 · arxiv updated 2021/04/26 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28
An important pillar for safe machine learning (ML) is the systematic mitigation of weaknesses in neural networks to afford their deployment in critical applications. An ubiquitous class of safety risks are learned shortcuts, i.e. spurious correlations a network exploits for its decisions that have no semantic connection to the actual task. Networks relying on such shortcuts bear the risk of not generalizing well to unseen inputs. Explainability methods help to uncover such network vulnerabilities. However, many of these techniques are not directly applicable if access to the network is constrained, in so-called black-box setups. These setups are prevalent when using third-party ML components. To address this constraint, we present an approach to detect learned shortcuts using an interpretable-by-design network as a proxy to the black-box model of interest. Leveraging the proxy's guarantees on introspection we automatically extract candidates for learned shortcuts. Their transferability to the black box is validated in a systematic fashion. Concretely, as proxy model we choose a BagNet, which bases its decisions purely on local image patches. We demonstrate on the autonomous driving dataset A2D2 that extracted patch shortcuts significantly influence the black box model. By efficiently identifying such patch-based vulnerabilities, we contribute to safer ML models.