Probabilistic Gradient Coding via Structure-Preserving Sparsification PubID: pubid:

Yuxin Jiang, Wenqin Zhang, Lele Wang · 2026

Read Probabilistic Gradient Coding via Structure-Preserving Sparsification PubID: pubid: by Yuxin Jiang, Wenqin Zhang, Lele Wang (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-10374.txt.

Description

Jiang, Zhang, and Wang widen gradient coding to far more system sizes than past codes allow. Gradient coding guards shared sums against slow or dead worker nodes in split computing. The classic BIBD design gives the best worst case trade, yet fits only a narrow set of sizes. Two chance built codes keep its core traits: one guards the BIBD shape, one guards key spectral traits. Trials match BIBD worst case error where BIBD fits and stretch far past its range.

Catalog record

CREATORYuxin Jiang, Wenqin Zhang, Lele Wang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE93810
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.10374

Browse the full library · About the Tower of Babel