it’s owl-ML4Pro²: Maschinelles Lernen für die Produktion und deren Produkte

Motivation

The aim of the joint project is to make machine learning (ML) for intelligent technical systems (ITS) available on a sustainable basis. This requires the transfer of state-of-the-art ML methods to the key areas of action in ITS and to bring ML technologies into the products and into the production chains, and conversely, the awareness of companies, when and how ML can be integrated into agile business models and production chains. The joint project is based on the digitization strategies of the participating companies and the ML expertise of the participating research partners in order to realize a step towards the efficient use of digital data by ML.

Challenges

Technical innovations are increasingly based on machine learning. ML has the potential to generate added value through the extraction of knowledge from digital data at all stages of the company's processes. With the current ML research topics "Hybrid Learning Techniques", "Integration of Expert Knowledge", "Explainability" and "Learning on Data Streams in Embedded Systems", the collaborative project addresses key issues for ITS. ML methods are considered across applications on the basis of three industrial applications, which are trend-setting for both production and their products.

Research Activities

The applicability of ML processes in the industrial environment is based centrally on the interpretability of the models considered. For a broad acceptance of such methods, questions of the validation of ML methods must be clarified. This requires not only insight into the procedures, but also the integration of prior knowledge in order to generate semantically meaningful explanations.

Another aspect in this context is determining the credibility of an information source. It is important to distinguish whether the data provided by the measurement technology reflects the system state or indicates defective sensors and environmental influences. Furthermore, the quality assessment and subsequent feedback on influencing the process and expanding the knowledge base is a topic that receives insufficient attention in the context of ITS.