CAO: Curvature-Adaptive Optimization via Periodic Low-Rank Hessian Sketching

Fuente: arXiv
Saved in:
Bibliographic Details
Main Author: Du, Wenzhang
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866909905579933696
author Du, Wenzhang
author_facet Du, Wenzhang
contents First-order optimizers are reliable but slow in sharp, anisotropic regions. We study a curvature-adaptive method that periodically sketches a low-rank Hessian subspace via Hessian--vector products and preconditions gradients only in that subspace, leaving the orthogonal complement first-order. For L-smooth non-convex objectives, we recover the standard O(1/T) stationarity guarantee with a widened stable stepsize range; under a Polyak--Lojasiewicz (PL) condition with bounded residual curvature outside the sketch, the loss contracts at refresh steps. On CIFAR-10/100 with ResNet-18/34, the method enters the low-loss region substantially earlier: measured by epochs to a pre-declared train-loss threshold (0.75), it reaches the threshold 2.95x faster than Adam on CIFAR-100/ResNet-18, while matching final test accuracy. The approach is one-knob: performance is insensitive to the sketch rank k across {1,3,5}, and k=0 yields a principled curvature-free ablation. We release anonymized logs and scripts that regenerate all figures and tables.
format Preprint
id arxiv_https___arxiv_org_abs_2511_12548
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle CAO: Curvature-Adaptive Optimization via Periodic Low-Rank Hessian Sketching
Du, Wenzhang
Machine Learning
68T07, 90C26, 90C30, 65K05
I.2.6; G.1.6; I.5.1
First-order optimizers are reliable but slow in sharp, anisotropic regions. We study a curvature-adaptive method that periodically sketches a low-rank Hessian subspace via Hessian--vector products and preconditions gradients only in that subspace, leaving the orthogonal complement first-order. For L-smooth non-convex objectives, we recover the standard O(1/T) stationarity guarantee with a widened stable stepsize range; under a Polyak--Lojasiewicz (PL) condition with bounded residual curvature outside the sketch, the loss contracts at refresh steps. On CIFAR-10/100 with ResNet-18/34, the method enters the low-loss region substantially earlier: measured by epochs to a pre-declared train-loss threshold (0.75), it reaches the threshold 2.95x faster than Adam on CIFAR-100/ResNet-18, while matching final test accuracy. The approach is one-knob: performance is insensitive to the sketch rank k across {1,3,5}, and k=0 yields a principled curvature-free ablation. We release anonymized logs and scripts that regenerate all figures and tables.
title CAO: Curvature-Adaptive Optimization via Periodic Low-Rank Hessian Sketching
topic Machine Learning
68T07, 90C26, 90C30, 65K05
I.2.6; G.1.6; I.5.1
url https://arxiv.org/abs/2511.12548