Generative/Stochastic/Bayesian Space-Time Neural Networks

Ilze Amanda Auzina (Ph.D. Student)

The PhD project explores the intersection between machine learning and dynamical systems. Throughout the project multiple research directions are investigated. Firstly, the effect of constraining networks with known dynamical properties is researched. Secondly, the project explores dynamical forecasting when the underlying dynamical system is unknown, however, one can identify underlying static factors of variation. Furthermore, the project looks into disentangling and identifying dynamic factors via interventions and causal modeling. Lastly, the project investigates the concept of dynamics in the context of state-of-the-art machine learning models, such as transformers and diffusion models.

Primary Host:	Efstratios Gavves (University of Amsterdam)
Exchange Host:	Matthias Bethge (University of Tübingen)
PhD Duration:	01 October 2021 - 01 October 2025
Exchange Duration:	01 February 2025 - 01 September 2025 - Ongoing