Four Hessian-informed trust-region filter variants using low- and high-fidelity surrogates reduce iterations and black-box evaluations by up to an order of magnitude on 25 benchmarks and five engineering cases while lowering tuning sensitivity.
Trust -region algorithms for training responses: machine learning methods using indefinite Hessian approximations
8 Pith papers cite this work. Polarity classification is still indexing.
8
Pith papers citing it
citation-role summary
background 1
citation-polarity summary
verdicts
UNVERDICTED 8roles
background 1polarities
background 1representative citing papers
ML-SPnP accelerates stochastic PnP for SVCT by using MRA approximation spaces where prior-coherence corrections vanish in expectation, yielding comparable quality at reduced runtime.
Reformulates SDCMPCC via spectral decomposition of complementarity structure and proves augmented Lagrangian accumulation points are W-stationary (or C-stationary under stricter subproblem conditions).
A general framework for parameter-free smooth nonconvex optimization via higher-order regularization yields algorithms with optimal complexity bounds without prior parameter knowledge.
Approximate directional stationarity is formulated as a necessary optimality condition for nonsmooth constrained problems, with a qualification condition using one sequence to infer directional stationarity.
Sufficient conditions on eigenvalue vanishing in quasi-Newton updates, observed numerically, are shown to imply convergence to criticality for piecewise differentiable nonsmooth functions, along with the method's ability to explore piecewise structure.
Hybrid GBS with classical post-processing for DkSP achieves near-optimal solutions and ~4X sampling efficiency gains on community graphs while outperforming pure post-selection on sparse graphs.
A solver-agnostic condensing reformulation for linear-quadratic optimization with polyhedral and geometric constraints that preserves augmented-Lagrangian convergence while improving computational speed.
citing papers explorer
-
Trust-region filter algorithms utilizing Hessian information for gray-box optimization
Four Hessian-informed trust-region filter variants using low- and high-fidelity surrogates reduce iterations and black-box evaluations by up to an order of magnitude on 25 benchmarks and five engineering cases while lowering tuning sensitivity.
-
Multilevel Stochastic Plug-and-Play for Sparse-View CT Reconstruction
ML-SPnP accelerates stochastic PnP for SVCT by using MRA approximation spaces where prior-coherence corrections vanish in expectation, yielding comparable quality at reduced runtime.
-
Augmented Lagrangian methods for nonlinear semidefinite programming with complementarity constraints
Reformulates SDCMPCC via spectral decomposition of complementarity structure and proves augmented Lagrangian accumulation points are W-stationary (or C-stationary under stricter subproblem conditions).
-
A General Recipe for Parameter-Free Nonconvex Optimization via Higher-Order Regularization
A general framework for parameter-free smooth nonconvex optimization via higher-order regularization yields algorithms with optimal complexity bounds without prior parameter knowledge.
-
Approximate directional stationarity and associated qualification conditions
Approximate directional stationarity is formulated as a necessary optimality condition for nonsmooth constrained problems, with a qualification condition using one sequence to infer directional stationarity.
-
Technical results on the convergence of quasi-Newton methods for nonsmooth optimization
Sufficient conditions on eigenvalue vanishing in quasi-Newton updates, observed numerically, are shown to imply convergence to criticality for piecewise differentiable nonsmooth functions, along with the method's ability to explore piecewise structure.
-
Towards a Hybrid Quantum Enhanced Solution for Densest k-Subgraph Problem
Hybrid GBS with classical post-processing for DkSP achieves near-optimal solutions and ~4X sampling efficiency gains on community graphs while outperforming pure post-selection on sparse graphs.
-
A condensing approach for linear-quadratic optimization with geometric constraints
A solver-agnostic condensing reformulation for linear-quadratic optimization with polyhedral and geometric constraints that preserves augmented-Lagrangian convergence while improving computational speed.