Semantic, Symbolic and Interpretable Machine Learning

This program concerns machine learning (ML) approaches which operate at the human abstraction level, where the world is described by entities, concepts, and their mutual relationships.  Among other topics, we cover multi-relational learning, learning with (temporal) knowledge graphs, and the extraction of statistical and logical regularities from data.  Methods include, e.g., embedding approaches, graph neural networks, scene graph analysis, neuro-symbolic programming, and inductive logic programming.

The goal is that our program becomes a cumulation point of like-minded researchers and we expect fruitful interactions with closely related programs covering, e.g., NLP, vision, and geometric deep learning.