Machine Learning

In Machine Learning (ML), algorithms generate predictive models from sample data. These models represent the knowledge acquired and can be applied to new, unknown data of the same type. Machine Learning is particularly helpful for correlations that are difficult to describe analytically but for which there is a large amount of sample data, such as sensor data, images, or text.

For companies, Machine Learning offers the advantage of automatically recognizing complex relationships even in large amounts of data and gaining valuable insights from them. This enables better forecasts, individual recommendations, and data-based decisions.

For many years, we have maintained a leading position in applied ML research. From comprehensible feature-based models to deep learning from unstructured data to foundation models of generative AI. We master and combine the most suitable approaches for designing human-centered, comprehensible, and trustworthy solutions.

We also work on applications that are resource-efficient and run on a wide variety of hardware platforms. We adapt our methods to individual application scenarios, from distributed learning (edge computing) to ML on quantum computers. Our learning algorithms thus open up previously inaccessible areas of application for business and society. Our solutions are used, for example, in the financial and insurance industries, control industrial production and quality control, and optimize administrative processes and procedures.

In studies and scientific publications, we report on the potential and limitations of these technologies. We are involved in key research projects in Germany and Europe and contribute to the social debate on the impact of intelligent machines.