June 22, 2022

Saarbrücken AI research celebrates international successes

Bild der Pressemitteilung


Saarland University computer science professor Jörg Hoffmann has achieved several successes with his research group “Foundations of Artificial Intelligence” at a leading international conference in a subfield of AI research, the “International Conference on Automated Planning and Scheduling (ICAPS)”. Former doctoral student Daniel Gnad wins the conference’s dissertation award, and an exceptionally large proportion of the research papers accepted at the conference was co-authored by the group.

Jörg Hoffmann is professor of computer science at Saarland University since 2012. The artificial intelligence expert focuses on action-decision mechanisms in complex planning problems, such as chess or production planning. Recently his focus is on trustworthy AI, specifically methods for quality assurance of machine-learned decision processes.

The “International Conference on Automated Planning and Scheduling (ICAPS)” is one of the leading scientific conferences in this subfield of AI research. For the 2022 edition, about 70 research groups from around the globe submitted hundreds of research papers for discussion at the conference, of which 85 papers were accepted. As Jörg Hoffmann explains: “My research group achieved a spectacular success: Of the 85 accepted papers, 11 have co-authorship from my research group. This is a new record – in the more than 30-year history of the conference series, no research group has ever had more than 6 papers accepted at an edition of ICAPS.”

Furthermore, Daniel Gnad’s PhD thesis, conducted in Jörg Hoffmann’s group, won the “Best Dissertation Award”. The thesis, entitled “Star-Topology Decoupled State-Space Search in AI Planning and Model Checking”, contributes a novel method for efficient planning that leverages independence properties, and achieved outstanding results also in safety verification.

Professor Jörg Hoffmann already received several honors for his contributions to AI last year: At the beginning of 2021, he was named a Fellow of the “Association for Artificial Intelligence (AAAI)”, making him one of only five scientists working in Germany to have received this accolade since 1990. Later last year, he was also appointed Fellow of the “European Association for Artificial Intelligence (EurAI)”.

 

ICAPS-2022 Accepted Papers (co)-authored by FAI

Daniel Höller, Julia Wichlacz, Pascal Bercher and Gregor Behnke. Compiling HTN Plan Verification Problems into HTN Planning Problems

Marcel Vinzent, Marcel Steinmetz and Jörg Hoffmann. Neural Network Action Policy Verification via Predicate Abstraction

Daniel Höller and Gregor Behnke. Encoding Lifted Classical Planning in Propositional Logic

Patrick Ferber, Florian Geißer, Felipe Trevizan, Malte Helmert and Joerg Hoffmann. Neural Network Heuristic Functions for Classical Planning: Bootstrapping and Comparison to Other Methods

Marcel Steinmetz, Daniel Fišer, Hasan Ferit Enişer, Patrick Ferber, Timo Gros, Philippe Heim, Daniel Höller, Xandra Schuler, Valentin Wüstholz, Maria Christakis and Joerg Hoffmann. Debugging a Policy: Automatic Action-Policy Testing in AI Planning

Martim Brandao, Amanda Coles, Andrew Coles and Joerg Hoffmann. Merge and Shrink Abstractions for Temporal Planning

Rebecca Eifler, Martim Brandao, Amanda Coles, Jeremy Frank and Joerg Hoffmann. Evaluating Plan-Property Dependencies: A Web-Based Platform and User Study

Thorsten Klößner, Marcel Steinmetz, Álvaro Torralba and Jörg Hoffmann. Pattern Selection Strategies for Pattern Databases in Probabilistic Planning

Daniel Gnad, Álvaro Torralba and Daniel Fišer. Beyond Stars – Generalized Topologies for Decoupled Search

Daniel Fišer, Álvaro Torralba and Joerg Hoffmann. Operator-Potentials in Symbolic Search: From Forward to Bi-Directional Search

Thorsten Klößner, Florian Pommerening, Thomas Keller and Gabriele Röger. Cost Partitioning Heuristics for Stochastic Shortest Path Problems

Further Information

https://icaps22.icaps-conference.org/
http://fai.cs.uni-saarland.de/research.html

Award-winning PhD-Thesis:
D. Gnad, Star-Topology Decoupled State-Space Search in AI Planning and Model Checking, Saarland University, 2021.

 

Background Saarland Informatics Campus:
900 scientists (including 400 PhD students) and about 2100 students from more than 80 nations make the Saarland Informatics Campus (SIC) one of the leading locations for computer science in Germany and Europe. Five world-renowned research institutes, namely the German Research Center for Artificial Intelligence (DFKI), the Max Planck Institute for Informatics, the Max Planck Institute for Software Systems, the Center for Bioinformatics and the Cluster for “Multimodal Computing and Interaction” as well as Saarland University with three departments and 24 degree programs cover the entire spectrum of computer science.

Editor:
Philipp Zapf-Schramm
Saarland Informatics Campus
Phone: +49 681 302-70741
E-Mail: pzapf@mmci.uni-saarland.de