We are happy to announce our workshop “Linear Operators in AI”.
The worlds of scientific computing and artificial intelligence are mostly developing in parallel. While "classical" scientific methods have strong mathematical foundations, they come with practical limitations, for example, in higher dimensions. AI methods are often used in such contexts, but they lack theoretical results on convergence and generalization. The workshop "Linear Operators for AI" aims to reduce this gap and focuses on the connection between scientific computing and AI through linear operators. Specifically, the workshop will consider two interconnected topics: understanding neural networks through operators and learning operators with neural networks.
The first topic covers, among others, bounding the generalization properties of neural networks with Gaussian processes and learning data geometry with the Laplace-Beltrami operator. The second topic considers operators emerging in dynamical systems, such as the Koopman operator and the solution operator for PDEs. We invite speakers to present reliable and scalable neural surrogate models for such operators.
The workshop will take place on 24-25 July 2025 (Thursday-Friday) at the Technical University of Munich, Germany. The workshop program will start at noon on Thursday and include invited talks and discussion sessions on the future of AI-driven scientific computing. We expect 30 to 40 participants and will provide funding for invited speakers. There is also the possibility of applying for financial aid for other participants.
Invited speakers (the list will be updated):
- Erik Bolager, Technical University of Munich (Germany)
- Tyrus Berry, George Mason University (USA)
- Nadiia Derevianko, Technical University of Munich (Germany)
- Debarghya Ghoshdastidar, Technical University of Munich (Germany)
- Stefan Klus, Heriot-Watt University (UK)
- Igor Mezic, University of California (USA), virtually
- Zahra Monfared, Heidelberg University (Germany)
- Feliks Nüske, Max Planck Institute (Germany)
Information about registration for participants will appear here soon.