Unifying AMP Algorithms for Rotationally-Invariant Models

Songbin Liu Junjie Ma · 2024

Read Unifying AMP Algorithms for Rotationally-Invariant Models by Songbin Liu Junjie Ma (2024) 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-2412-01574.txt.

Description

Songbin Liu and Junjie Ma present a math paper from 2024 on a family of methods called approximate message passing. These methods solve hard estimation tasks step by step, such as pulling a weak signal out of noisy data. The paper builds one shared frame for versions of the method used when the data matrix looks the same from every angle.

Starting from a general step-by-step scheme, the authors shrink it down to a known long-memory form. That route lets them work out the correction terms each method needs, and it brings back a known method from earlier work in a new light. A set of values called free cumulants turns out to drive the corrections. The frame also yields two fresh variants, which the authors test on spiked models.

Catalog record

CREATORSongbin Liu Junjie Ma
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE154 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-01574

Browse the full library · About the Tower of Babel