Machine Learning
with Graphs
Created using Midjourney
Research
Graphs and networks are ubiquitous in various domains, from chem- and bioinformatics to computer vision and social network analysis. Machine learning with graphs aims to exploit the potential of the growing amount of structured data in all these areas to automate, accelerate, and improve decision-making. Analyzing graph data requires solving problems at the boundaries of machine learning, graph theory, and algorithmics.
Our ambition is to develop methods that are useful for solving concrete problems in real-world applications, especially in drug discovery. We focus on the development of new machine learning and data mining methods for structured data related to the following broad topics:
Graph Embedding
Graph embedding techniques map graphs into a vector space, such that similar graphs are represented by close vectors. We investigate graph kernels and graph neural networks (GNNs) for embedding graphs to solve learning problems. While graph kernels allow operating implicitly in high-dimensional feature spaces, graph neural networks optimize the feature representation for the learning task. We design, theoretically analyze, implement, and empirically evaluate graph embedding methods.
Graph Matching
The term graph matching refers to methods for finding a correspondence between the nodes of two graphs that optimally preserves their node/edge features and adjacency structure. We study exact polynomial-time algorithms for restricted graph classes, heuristics for general large graphs, and learning-based graph matching techniques.
Graph Search
Information systems with graph-structured data require custom methods for specifying, identifying, and retrieving the desired information. We develop efficient methods and index structures for searching large databases of graphs, e.g., regarding similarity or subgraph containment.
Team
Nils
Kriege
Group Leader
Franka
Bause
PhD Candidate
Lorenz Kummer
PhD Candidate
Anatol
Ehrlich
PhD Candidate
Tinatini Buturishvili
External PhD Candidate
Ole
Schlüter
Research Assistant
Charlotte
Zott
Administration
News
-
Kriege Group Website Goes Online!
–
The kriegegroup.univie.ac.at website has officially launched on 11th November 2024. Its aim is to consolidate the information about the Machine Learning with Graphs work group to make our research more accessible to interested individuals. For the more tech savvy people, here is a quick overview of the tech stack used on this website: WordPress as…
-
Kurier “Wien will’s wissen” features research by Prof. Nils Kriege
–
Professor Nils Kriege was featured in an article discussing how Artificial Intelligence assists in drug discovery by analyzing molecular data to identify new medications and the potential this technology offers for further development and research. For more details, you can read the article online on futurezone.at
-
Franka presents at ECML/PKDD 2024
–
The paper Approximating the Graph Edit Distance with Compact Neighborhood Representations by Franka Bause (DoCSunivie student, Cluster D&K), Christian Permann and Nils M. Kriege (DoCSunivie Supervisor, Cluster D&K) was accepted at ECML/PKDD 2024. You can see the full version of the paper at Springer Link. The European Conference on Machine Learning and Principles and Practice…
Join Us!
We are a collaborative and ambitious research team advancing machine learning with graphs through both theoretical analysis and methodological innovation for applications.
PhD Students
We regularly have open positions for doctoral researchers. We are open to working with outstanding students from different areas, including computer science, and mathematics. If you are interested, please send your CV and up-to-date transcript of records to Nils Kriege.
Master’s and Bachelor’s Students
We are happy to supervise students at the University of Vienna who are interested in our research. An overview of possible topics for theses offered by our group can be found in the teaching section or on the Univie Open Topics page. Please get in touch with Nils Kriege or the specified contact persons and send an up-to-date transcript of records.