Marios Koulakis

I work on deep learning theory, data and representation geometry, manifold learning, dimensionality reduction, clustering, mechanistic interpretability, and trustworthy AI, with a focus on geometric benchmarks for learned representations.