Modern data can be seen as a giant web of interconnected variables — like a social network of numbers. By learning the hidden patterns in data (aka structure), we can understand out how the system behind it really works. This knowledge is the driving force behind AI, self-driving cars, disease cures, climate models, Spotify playlists, and countless other discoveries.
The challenge is that data is often messy — noisy, incomplete, biased, scarce — so classical methods like PCA are not enough. We create the next generation of methods that can uncover patterns hidden in messy data.