Research

The common thread is representation learning and verification: what a model has actually learned, and whether we can check it. The subjects range from formal proofs of code to markets, touch and cancer genomics. Grey marks are the other five layers, kept in place so positions stay honest.