Vortragsankündigung: Machine Learning in Formal Logics on Knowledge Graphs
Logic is the traditional scientific approach to represent knowledge in a formal way and draw conclusions from given knowledge. Machine learning, on the other hand, targets similar challenges and draws early roots from logical frameworks. In this talk, we will explore the evolution of machine learning from logic-based approaches to modern parameter-based models and introduce new initiatives in logic-based machine learning. Limitations of modern parameterbased methods, such as the lack of explainability and efficiency, have motivated the renaissance of the logic-based approach. The recent development of largescale knowledge graphs and the semantic web initiatives further provide rich sources of structured data to be exploited in logic-based machine learning. We will conclude the talk with a brief introduction of the ENEXA project and the
current challenges of machine learning on knowledge graphs to be investigated within this project.