Generative AI is transforming everything, yet it still struggles with one of machine learning's oldest challenges: making accurate predictions from tabular data. For years, decision trees have dominated, but this is now beginning to change. A new wave of Tabular Foundation Models is emerging: models that learn not just from single datasets, but across many tasks, transferring both statistical patterns and world knowledge. In this talk, we'll introduce TabSTAR, a hybrid architecture built to fuse the strengths of deep learning with the unique structure of tabular data, and explain why and how this could spark the next big revolution in predictive AI.
Room: A4+A5
Tue, Oct 28th, 9:00 - 9:30