Saved in:
Bibliographic Details
Main Authors: Kabgani, Alireza, Lara, Felipe, Ahookhosh, Masoud
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2605.08474
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915995871870976
author Kabgani, Alireza
Lara, Felipe
Ahookhosh, Masoud
author_facet Kabgani, Alireza
Lara, Felipe
Ahookhosh, Masoud
contents Robust learning aims to maintain model performance under noise, corruption, and distributional shifts, which are prevalent in modern machine learning applications. This work shows that examples of robust learning problems can be formulated as (strongly) quasar-convex optimization problems, which admit a benign landscape with no saddle points. We then propose HiPPA, an inexact high-order proximal-point method that employs a model-value gap to control the inexactness of subproblem solutions. Notably, we prove global convergence of HiPPA to global minima and establish that it attains a (local) linear or superlinear convergence rate, depending on the regularization order and inexactness control. Our numerical experiments on robust feature-alignment distillation indicate strong empirical performance of HiPPA and results consistent with our theoretical findings.
format Preprint
id arxiv_https___arxiv_org_abs_2605_08474
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Robust Learning Meets Quasar-Convex Optimization: Inexact High-Order Proximal-Point Methods
Kabgani, Alireza
Lara, Felipe
Ahookhosh, Masoud
Optimization and Control
Robust learning aims to maintain model performance under noise, corruption, and distributional shifts, which are prevalent in modern machine learning applications. This work shows that examples of robust learning problems can be formulated as (strongly) quasar-convex optimization problems, which admit a benign landscape with no saddle points. We then propose HiPPA, an inexact high-order proximal-point method that employs a model-value gap to control the inexactness of subproblem solutions. Notably, we prove global convergence of HiPPA to global minima and establish that it attains a (local) linear or superlinear convergence rate, depending on the regularization order and inexactness control. Our numerical experiments on robust feature-alignment distillation indicate strong empirical performance of HiPPA and results consistent with our theoretical findings.
title Robust Learning Meets Quasar-Convex Optimization: Inexact High-Order Proximal-Point Methods
topic Optimization and Control
url https://arxiv.org/abs/2605.08474