arXiv Machine Learning

Randomized and Oracle-Based Feasible Directions Methods for Nonsmooth Nonconvex Constrained Optimization

The paper introduces feasible‑direction methods for minimizing nonsmooth, nonconvex objectives over convex sets without using derivatives. It defines η‑feasible directions to overcome the non‑closedness of the feasible direction cone at boundary points, enabling finite‑direction search even with nonlinear constraints. Two algorithmic realizations are presented: one using an η‑direction oracle and another employing a noise‑augmented randomization scheme, both achieving finite‑time approximate d‑stationarity and establishing convergence rates under curvature and Lipschitz assumptions.

arXiv Machine Learning
Jul 9

Geometric--Nongeometric Optimizer Calculus: A Modular Language for Reachable Gradient Methods

arXiv:2607. 07206v1 Announce Type: new Abstract: Adaptive optimizers mix several mechanisms: a metric or preconditioner maps gradients to descent directions, while estimation, memory, step-size control, constraints, stochasticity, target modification, and discretization determine which directions are available and how they are used.

By Zavier Li
arXiv Machine Learning
Jul 23

Online Optimization of Difference-of-Convex Compositions with Smooth Mappings

arXiv:2607. 19553v1 Announce Type: cross Abstract: We study online optimization for a broad class of structured non-convex non-smooth problems where each loss is a composition of a difference-of-convex function with a smooth mapping, and the feasible region is defined by constraint functions of the same kind.

By Jingwei Ji, Jong-Shi Pang, Renyuan Xu
arXiv Machine Learning
Sep 30

Global Optimality for Constrained Exploration via Penalty Regularization

The paper introduces Policy Gradient Penalty (PGP), a single‑loop policy‑space method that enforces convex occupancy‑measure constraints via quadratic‑penalty regularization. PGP constructs pseudo‑rewards to estimate gradients of the penalized objective and uses the classical Policy Gradient Theorem, establishing smoothness and global last‑iterate convergence guarantees for an ε‑optimal constrained entropy value with ε‑bounded constraint violation. The authors validate PGP with ablations on a grid‑world benchmark and demonstrate scalability on two challenging continuous‑control tasks.

By Florian Wolf, Ilyas Fatkhullin, Niao He
arXiv Machine Learning
Oct 5

Learn Feasibility Once, Optimize All Objectives: Derivative-Free Diffusion Models for Chance-Constrained Programming

The paper introduces D$^3$Opt, a derivative‑free diffusion‑based framework that separates constraint modeling from objective optimization in chance‑constrained programs. It learns a risk‑conditioned diffusion model once on constraint‑filtered decisions, then uses an annealed, particle‑based Feynman–Kac correction to optimize any post‑specified objective without retraining. The method is proven to preserve feasibility and provides an error bound, and experiments show effective optimization across smooth, non‑smooth, and non‑convex objectives, including objective transfer tasks.

By Ziwen Liu, Yan Liu, Congying Han, Tiande Guo, Yao Yan, Weichen Zhao