About:

Lifting high-dimensional problems to an infinite-dimensional space and designing algorithms in that setting has been a fruitful idea in many areas of applied mathematics, including inverse problems, optimisation, and partial differential equations. This approach is sometimes referred to as "optimise-then-discretise" and allows the development of algorithms that are inherently dimension- and discretisation-independent and can perform better in high-dimensions. In the context of machine learning, this paradigm can be rephrased as "learn-then-discretise".
This workshop aimed to bring together researchers who work on different aspects of infinite-dimensionality in machine learning. Topics include, but are not restricted to, Gaussian process regression, operator learning, function spaces of neural networks, and measure transport.

We acknowledge generous support of the International Centre for Mathematical Sciences, Edinburgh, the London Mathematical Society, the Prob AI Hub on the Mathematical and Computational Foundations of AI, and the EPSRC programme grant Maths4DL.

More details, including the registration form, can be found on the workshop's webpage https://sites.google.com/view/machine-learning-workshop-bath