Derives O(s^{1.5}/√N) generalization bound, Ω(s/√N) minimax lower bound, and shows SGA with averaging attains Θ(s/√N) optimal rate for data-driven Lagrangian relaxation in MILPs, plus faster Θ(s/N) rate for warm-start learning.
Title resolution pending
2 Pith papers cite this work. Polarity classification is still indexing.
2
Pith papers citing it
years
2026 2representative citing papers
W-SparQ-BL models time-varying lower-level responses with multi-output GPs and sparse approximations to achieve sublinear dynamic regret in bilevel optimization under noise.
citing papers explorer
-
Provably Data-driven Lagrangian Relaxation for Mixed Integer Linear Programming
Derives O(s^{1.5}/√N) generalization bound, Ω(s/√N) minimax lower bound, and shows SGA with averaging attains Θ(s/√N) optimal rate for data-driven Lagrangian relaxation in MILPs, plus faster Θ(s/N) rate for warm-start learning.
-
No-regret optimization of time-varying bilevel problems
W-SparQ-BL models time-varying lower-level responses with multi-output GPs and sparse approximations to achieve sublinear dynamic regret in bilevel optimization under noise.