In this course we introduce you to a variety of machine and deep learning methodology to process graph-structured data. We define graph-structured data to refer to the combination of an underlying graph (or network) structure on which vectorial data is observed at the nodes, edges or both. This data type is frequently observed in practice and hence a multitude of methods have been defined to learn from it. In this course, we will review fundamental summary statistics of graphs and probabilistic models to generate graphs; we will introduce you to graph kernel methods and then move on to provide you a comprehensive overview of deep learning methodology, notably Graph Neural Networks among others. We end the course with an review of applications of the introduced methodology and an outlook on current challenges and future directions in the domain of machine and deep learning on graph structured data.
The class will be evaluated with a combination of quizzes and team projects.
The course will be taught in English.
