Doctoral research · Leeds Beckett University · 2025 – present
Causal AI for decisions in the real world
Most AI systems are good at prediction but say little about cause and effect, which is what decision-makers need when they ask what would happen if they intervened.
My PhD develops causal AI methods, with a focus on causal discovery: learning cause-and-effect structure from observational data that is short, noisy, incomplete or changing over time. The methods are designed to be domain-agnostic, with food security systems as the use case for validation, where stronger causal evidence could support earlier and better-targeted interventions.
Themes
- Causal discovery from time series
- Reliable methods under distribution shift and missing data
- Moving from forecasts to interventions