Henok T. Moges

AI for Dynamical Systems & Scientific Machine Learning

University of Cape Town

"When does physical structure help learning in dynamical systems and when does it fail?"

I am a Research Fellow in AI Systems at the University of Cape Town, working at the intersection of machine learning, numerical analysis, and nonlinear dynamics.

My research focuses on spatio-temporal graph neural networks (STGNNs), physics-informed machine learning, and scientific ML for predictive modelling of complex dynamical systems. I study how learning architectures interact with nonlinear instability in high-dimensional systems, with emphasis on predictability horizons, stability, and chaos.

Occasionally, I try to understand chaotic systems. They mostly ignore me :)

Research Themes

Spatio-temporal Learning
Physics-informed Machine Learning
Nonlinear Dynamics & Chaos
Predictability in Complex Systems
Scientific ML & Benchmarking

Current Position

Research Fellow
Artificial Intelligence Research Unit (AIRU), University of Cape Town
Work Package Lead — AI & Modelling
Mi-Hy EU Project
AI-driven modelling, simulation, and digital twin design for microbial hydroponics systems

Selected Projects

ChaosBench

Benchmark for evaluating forecasting models under chaotic dynamical systems, focusing on predictability limits and evaluation under instability.

Lite-STGNN

Efficient spatio-temporal graph neural networks for long-term time series forecasting under resource constraints.

Physics-informed STGNNs

Hybrid models combining physical structure with deep learning for dynamical prediction.

Selected Publications

Full list on Google Scholar

Links