The invention relates to a method for evaluating and certifying the robustness (25) of an AI-based information processing system (10). At least one multidimensional data structure (20) corresponding to the AI-based information processing system (10) is received or generated, and respective differential values (22), determined at least using at least one differential measurement (13a), between output data of the AI-based information processing system (10), said output data being obtained for data (11a) and augmented data, are stored in the at least one multidimensional data structure (20) on the basis of at least the dimensions of the data set (11), the data augmentation definition(s) (12), and the differential measurement definition(s), wherein at least the robustness (25) of the AI-based information processing system (10) is determined from at least a selection of the differential values (22), and the robustness is compared with at least one robustness requirement (26). On the basis of the comparison result (29), the AI-based information processing system (10) is either rejected, evaluated again using a modified multidimensional data structure (20+), or certified as being robust. The invention additionally relates to a corresponding device (1).
展开▼
