3 ms·
Second order methods are attracted to saddle points in high dimensional spaces. The math and practice of optimizing these surfaces has a lot of nuances like thi
by jph00 9y ago
Second order methods are attracted to saddle points in high dimensional spaces. The math and practice of optimizing these surfaces has a lot of nuances like this so much of the stuff you learn in your convex optimization class doesn't apply too well.
- steev 9y agoDo you have any recommendations on sources to read about this? Everything I've read discusses the use of the Hessian to not only determine you are at a saddle point but to also use its eigenvalues to escape.