Collaborative Compressors in Distributed Mean Estimation with Limited Communication Budget

Harsh Vardhan [email protected] · 2026

Read Collaborative Compressors in Distributed Mean Estimation with Limited Communication Budget by Harsh Vardhan [email protected] (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-2601-18950.txt.

Description

Harsh Vardhan studies how many computers can agree on an average while sharing very little. In spread out learning, many machines each hold a vector, and the group needs the mean. Bandwidth is tight, so each vector must be squeezed before it is sent.

The twist here is teamwork: the squeezers, called compressors, can work together rather than each acting alone. Shared plans for squeezing can cut the error of the mean for the same small budget of bits. The result speaks to big training jobs where talk between machines, not math, is the main cost.

Catalog record

CREATORHarsh Vardhan [email protected]
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE94 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-18950

Browse the full library · About the Tower of Babel