ConceptHypergradient
Hypergradient
In bilevel optimization, the hypergradient is the derivative of the upper-level objective with respect to hyperparameters. It is typically computed via the implicit function theorem and depends on the inverse of the lower-level Hessian matrix.
Timeline (last 90 days)
Oct 20
Traditional Newton-type bilevel methods rely on the classical hypergradient formula, which fails when the lower-level problem's solution is non-unique or the lower-level Hessian is singular.
Unverified50%