AI.info

Emanuele Troiani

Explore Emanuele Troiani on AI.info.

• A solvable high-dimensional model where nonlinear autoencoders learn structure invisible to PCA while test loss misaligns with generalization