Machine learning is an applied research field

Professorship for Applied Computer Science,
especially cognitive systems

Brief portrait - More than 15 years of research on understandable machine learning

In the research field Cognitive Systems (CogSys) we deal with the development of approaches, concepts and methods for the design, characterization, implementation and analysis of artificial intelligence systems based on cognitive principles. On the one hand, we use knowledge from cognitive processes as a stimulus for the development of artificial (psychonic) systems. On the other hand, we develop computational models of cognitive phenomena - i.e. cognitive systems - that enable human and computer to interact as partners. In our research we combine empirical studies, the development of intelligent algorithms and testing in various application areas. The main topics are induction and learning and their combination with approaches of analog reasoning as well as planning and problem-solving. The focus is on inductive programming, i.e. the inductive synthesis of (recursive) functional or logical programs based on incomplete specifications, especially examples. The approaches we have developed make it possible to learn complex, comprehensible rules from a small amount of data. We have been researching understandable and explainable artificial intelligence for more than fifteen years. Here we develop white-box approaches to machine learning, in which learning can be combined with knowledge-based approaches. In addition, we develop methods for generating verbal and visual explanations of learned classifiers, especially for black box approaches to image classification. Current applications are the identification of irrelevant digital objects, quality control in various applications, in particular Industry 4.0, facial expression analysis, in particular pain classification, and cognitive tutor systems.