Research
This is what I'm currently working on.
I study how to build models that stay reliable when the data changes across environments.
In particular, I focus on causal regularization under distribution shifts, aiming to obtain stable and robust predictions even when we do not observe a reference setting.