Advances and fundamentals in ML

As in any science, fertility in Machine Learning also comes from cross-pollination. Keeping an eye open for interesting development across as many of the fields in ML as possible, we hope to identify developments with great potential. This includes techniques in optimisation, model evaluation and selection, function approximation and many other topics. At the same time, many fundamental concepts in ML are subtle, require careful analysis and must be often revisited and reexamined for new insights and developments.

Series in this AoI

Simulation and AI

AI techniques are fundamentally transforming the field of simulation by combining physics-based modeling with data-driven machine learning.

Geometric deep learning

Specialized deep learning architectures exploit the intrinsic regularities arising from the underlying structure of the physical world. Geometric deep learning aims to expose these regularities through unified geometric principles. Doing this allows us to understand why some deep learning architectures are especially successful and gives a constructive procedure to incorporate prior physical …

Diffusion Models

Diffusion models (DM) have become the state of the art for sample quality in generative modelling. They work by sequentially corrupting training data with slowly increasing noise and then learning to reverse the corruption. A look at their mathematical foundations reveals how much they borrow from statistical mechanics. By understanding their foundations and recent implementations, practitioners …

Reinforcement Learning

Recent and popular advances in Reinforcement Learning are known to be data-hungry. Attempts to handle this deficit include developing complex simulators while improving the sim2real transition of models, explicitly modelling the environment, or leveraging available offline data.