ExPro – Cooperative exploration environment of machine learning forecasts using the example of production planning
Duration: 2021-2023
Funded by: Federal Ministry of Education and Research
Contact person: Dr. Thomas Ludwig – Marco Fries – Julia Nießner
While AI procedures are already successfully used in cyber-physical assistance systems, it is almost completely lacking in the area of business organisation, although despite the availability of numerous company data are available that can selectively support human assessments with the help of e.g., machine learning (ML) methods. In everyday business, an explanatory component is missing so that the results of machine learning methods can be interpreted without additional qualifications. Methods from the field of explainable artificial intelligence (XAI) can remedy this.
The aim of the project is to develop a machine-learning module for forecasting production-relevant key figures that is based on historical company data and production plans and is comprehensible for end users. As an innovative core element, the application demonstrator enables a heterogeneous decision-making group and appropriation infrastructure to understand and justify machine-learning forecasts by exploring data sets with XAI methods in order to jointly reach a decision. This is to be realised and evaluated in the project in close cooperation with companies from the production environment and in the specific context of production planning.
Based on an analysis of the available data and user requirements of two manufacturing companies, an XAI framework for structured data and a collaborative working environment for the acquisition and exploration of ML models based on a low-code development platform will be developed in close consultation with the users. The final evaluation will take ethical, legal, and social aspects into account.
The intelligent evaluation of company data provides enormous potential for the business development of manufacturing companies. The project will create new opportunities for companies to use AI techniques in their departments without having to hire their own AI experts. The value network of research organisations, companies and social partners can develop and market a standardised service from this. For AI research, new opportunities for the further development of ML and XAI solution approaches arise due to the practical trials.
Further information will follow soon.
