Moritz Gögl

I am a DPhil student working on probabilistic methods in machine learning, supervised by Christopher Yau. I am interested in how models can learn meaningful representations, quantify uncertainty, and make reliable predictions from complex and structured data.

My work investigates methodological questions in machine learning from a probabilistic perspective, across topics including self-supervised learning, causal inference, language models, and survival analysis. In self-supervised learning, I study predictive representation learning and variational formulations of joint-embedding predictive architectures (JEPAs). I also investigate probabilistic approaches for causal effect estimation and survival analysis, with applications to personalised prediction and multimodal clinical data.