[D] Hessian of deep learning model and its eigenvector

I'm trying to understand more about Hessian matrix, its eigenvector, and how to optimize them. Can anyone provide me some insight about this (or recommend papers that has great insight about this aspect)? Something like what eigenvectors of a Hessian matrix represent, how they should behave in a good and generalized model.

deep learning eigenvectors generalized good insight machinelearning matrix papers something them