Papers › Perturbated Gradients Updating within Unit Space for Deep Learning

Perturbated Gradients Updating within Unit Space for Deep Learning

1 Oct 2021arXiv:2110.00199archive 2025-07-28

Ching-Hsun. Tseng, Liu-Hsueh. Cheng, Shin-Jye. Lee, Xiaojun Zeng

In deep learning, optimization plays a vital role. By focusing on image classification, this work investigates the pros and cons of the widely used optimizers, and proposes a new optimizer: Perturbated Unit Gradient Descent (PUGD) algorithm with extending normalized gradient operation in tensor within perturbation to update in unit space. Via a set of experiments and analyses, we show that PUGD is locally bounded updating, which means the updating from time to time is controlled. On the other hand, PUGD can push models to a flat minimum, where the error remains approximately constant, not only because of the nature of avoiding stationary points in gradient normalization but also by scanning sharpness in the unit ball. From a series of rigorous experiments, PUGD helps models to gain a state-of-the-art Top-1 accuracy in Tiny ImageNet and competitive performances in CIFAR- {10, 100}. We open-source our code at link: https://github.com/hanktseng131415go/PUGD.

PaperPDFCode

In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.

Code

hanktseng131415go/pugd officialmentioned in paperpytorch report

Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.

Code Syntology ran Syntology

Not run by Syntology. Nothing on this page verifies that the listed code works.

Tasks

Deep LearningImage Classificationimage-classification

Results from the paper archive 2025-07-28

TaskDatasetModelMetricValueRank at snapshotLeaderboardReport
Image Classification CIFAR-10 ViT-B/16 (PUGD) Percentage correct 99.13 #12 of 265 Archive leaderboard report
Image Classification CIFAR-100 ViT-B/16 (PUGD) Percentage correct 93.95 #6 of 211 Archive leaderboard report
Image Classification Tiny ImageNet Classification DeiT-B/16 (PUGD) Validation Acc 91.02% #5 of 23 Archive leaderboard report
Image Classification Tiny ImageNet Classification ViT-B/16 (PUGD) Validation Acc 90.74% #7 of 23 Archive leaderboard report

Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.

Methods

Gradient NormalizationSharpness-Aware Minimization

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections