COPLAS talk: Wolfgang Maaß
Speaker
Wolfgang Maaß, with degrees from RWTH Aachen University and Saarland University, including a Ph.D. in Computer Science, is a distinguished professor at Saarland University, specializing in Business Informatics and Computer Science. He also serves as a Senior Data Science Advisor at the National Cancer Institute (NCI) in the United States. As the scientific director at the German Research Center for Artificial Intelligence (DFKI), Maaß focuses on research in data economics and applying artificial intelligence across various industries such as manufacturing and health.
Title
Quantum-Enhanced Graph Neural Networks for Scalable Manufacturing Simulations
Abstract
Complex real-world tasks, such as manufacturing simulations, weather forecasting, and economic predictions, demand high computational efficiency to optimize processes and minimize errors. In manufacturing, simulations rely on solving partial differential equations (PDEs) over mesh representations, traditionally addressed using Finite Element Methods (FEM) or, increasingly, Machine Learning (ML) approaches. However, FEM is computationally intensive, while classical ML models require substantial resources. This work explores the integration of quantum computing to enhance these models and improve efficiency.
Quantum Feature Embeddings (QFEs) leverage quantum circuits to encode node features, transforming input data into quantum states for non-linear and entangled representations. This approach reduces qubit requirements while maintaining superior performance. Benchmark evaluations show that QFEs outperform classical counterparts with the same parameter count and match the performance of exponentially larger classical models. Applying a hybrid quantum-classical Graph Neural Network (GNN) to laser cutting demonstrates strong industrial potential.
Building on these insights, Quantum-based Mesh-Graph Networks (QMGN) are introduced to accelerate manufacturing simulations by leveraging quantum advantages. A case study on heat flow simulation in laser cutting evaluates QMGN’s ability to predict temperature evolution. Results indicate that QMGN outperforms classical GNNs and significantly speeds up FEM-based simulations. Incorporating geometric distances between mesh nodes further enhances accuracy, highlighting the importance of spatial relationships.
To ensure scalability for larger and more complex graphs, QMGN integrates advanced techniques such as local subgraph modeling, neighborhood subsampling, and subgraph attention mechanisms. These findings underscore the potential of quantum-enhanced GNNs in industrial simulations, particularly in laser cutting, and their broader applicability in computationally demanding manufacturing tasks.
Host
Associate Professor Boris Düdder
If you have trouble finding the meeting room, please call Fritz Henglein at +45-30589576.
All are welcome. No registration required. Feel free to forward this invitation.
The Copenhagen Programming Languages Seminar (COPLAS) is a collaboration between DTU, ITU, Roskilde University and UCPH. To be informed about COPLAS activities and related talks, join this mailing list.