What are the mathematical principles that make modern artificial intelligence possible—and how can they help create the next generation of reliable machine learning algorithms? These questions will be explored at the Speinshart Scientific Center for AI and SuperTech during the international workshop “Multivariate Approximation Theory: From Classical Methods to Deep Learning.”
The workshop brings together internationally recognised experts in approximation theory, numerical analysis, machine learning, and artificial intelligence from leading universities in Europe, North America, Asia, and Australia. Their common goal is to bridge the gap between the mature mathematical theory of classical approximation methods and the rapidly evolving field of deep learning.
While neural networks have achieved remarkable success across countless applications—from image recognition and natural language processing to scientific computing and medicine—their mathematical foundations remain only partially understood. By contrast, classical approximation techniques such as kernel methods, radial basis functions, splines, and wavelets are supported by decades of rigorous theoretical research. The workshop seeks to combine these complementary perspectives in order to develop more accurate, reliable, and theoretically grounded machine learning methods.
Over three days, participants will discuss fundamental questions including the expressive power of deep neural networks, convergence and stability of learning algorithms, optimisation methods, high-dimensional approximation, and the mathematical challenges posed by increasingly complex AI architectures. Alongside plenary lectures and scientific discussions, the programme will provide ample opportunity for collaborative research, informal exchange, and a dedicated poster session for early-career researchers.
By bringing together leading mathematicians from diverse yet closely connected disciplines, the workshop aims to establish new long-term collaborations and advance the theoretical foundations of artificial intelligence. The Speinshart Scientific Center is delighted to host this international gathering, reaffirming its commitment to supporting fundamental research that underpins the future development of trustworthy, robust, and scientifically grounded AI technologies.
Please note: This is just an information regarding events taking place at SSC; public attendance is therefore not possible due to the character of the retreat.
