DIN SPEC 92001-2 - Artificial Intelligence - Life Cycle Processes and Quality Requirements - Part 2: Robustness

This DIN SPEC in accordance with the PAS procedure has been drawn up by a DIN SPEC (PAS)-consortium set up on a temporary basis. This DIN SPEC (PAS) has been developed and approved by the authors named in the foreword. The aim of this series of DIN SPECs is to enable the safe and transparent development and use of AI modules. To achieve this, the DIN SPEC describes a number of AI quality requirements which are structured using an AI quality meta model. The DIN SPEC series applies to all phases of the life cycle of an AI module (i.e. its conceptualization, development, use, operation and termination) and takes into account a number of different life cycle processes. AI technologies are used for a diverse range of applications, which is why this DIN SPEC series is aimed at companies across all sectors. Part 2 of the DIN SPEC 92001 series describes the quality requirements specific to AI robustness. These quality requirements are structured using the specified AI quality meta model (DIN SPEC 92001-1). DIN SPEC 92001-2 explains various mathematical AI robustness requirements, specifically regarding Adversarial Robustness (i.e. AI attacks with mathematically optimized perturbations leading to model failure) and Corruption Robustness (i.e. model sensitivities to naturally occurring noise / data outliers leading to model failure). © 2023 Beuth Verlag GmbH

The information about this standard has been compiled by the AI Standards Hub, an initiative dedicated to knowledge sharing, capacity building, research, and international collaboration in the field of AI standards. You can find more information and interactive community features related to this standard by visiting the Hub’s AI standards database here. To access the standard directly, please visit the developing organisation’s website.