Development of methods for incorporating knowledge into machine learning
AI-based processes pave the way to fully automated driving. So far, the development of the AI processes has been purely data-based. Enormous amounts of data are required for the training and validation of the AI functions, the collection and processing of which is very time-consuming and expensive. In addition to being dependent on extensive amounts of data, data-based AI processes have another weakness: they are still usually black box models, their decision-making cannot be directly traced.
In the research project KI Wissen, methods for the integration of existing knowledge into the data-driven AI functions of autonomous vehicles are developed and examined. The aim is to create a comprehensive ecosystem for integrating knowledge into training and securing AI functions. By combining conventional data-based AI processes with the knowledge-based methods developed in the project, the basis for training and validation of the AI functions is completely redefined: This now includes not only data, but information, i.e. data and knowledge. The development from data-based to information-based AI, carried out in the project, addresses the central challenges on the way to autonomous driving: the generalization of the AI to phenomena with a low data basis, the increase in the stability of the trained AI to disruptions in the data, the data efficiency, the plausibility check and the protection of AI-supported functions as well as the increase in functional quality.