A safety-filtered gradient flow that enforces the lower-level optimality condition solves bilevel problems in a single loop, with a relaxed variant that avoids matrix inversions.
Meta-learning with implicit gradients
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
math.OC 1years
2025 1verdicts
CONDITIONAL 1roles
background 1polarities
unclear 1representative citing papers
citing papers explorer
-
Safe Gradient Flow for Bilevel Optimization
A safety-filtered gradient flow that enforces the lower-level optimality condition solves bilevel problems in a single loop, with a relaxed variant that avoids matrix inversions.