A variational approach to dimension-free self-normalized concentration

Ben Chugg, Aaditya Ramdas · 2025

Read A variational approach to dimension-free self-normalized concentration by Ben Chugg, Aaditya Ramdas (2025) 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-2508-06483.txt.

Description

Ben Chugg and Aaditya Ramdas sharpen bounds on how far random vector sums can stray. Their 2025 paper treats self-normalized sums, where the spread scale comes from the data itself. A broad class of tail types is covered, from light tails to some heavy ones with no moment curve.

Their variational route recovers a famed bound in the light-tail case and widens it. A gap between matrix-based and condition-based bounds gets filled. Payoffs include a Bernstein-style bound for vectors under a moment rule, and the first size-free self-normalized version of that bound. Such bounds underwrite safe bandit and trial math where data arrive in streams.

Catalog record

CREATORBen Chugg, Aaditya Ramdas
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE100 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-06483

Browse the full library · About the Tower of Babel