ScaleSpecter: a frequency-aware multi-scale patch framework for robust physiological classification under non-stationarity

Zhouyang Xu, Hongwei Li, Wenchao Liu, Haifeng Li · 2026

Read ScaleSpecter: a frequency-aware multi-scale patch framework for robust physiological classification under non-stationarity by Zhouyang Xu, Hongwei Li, Wenchao Liu, Haifeng Li (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/PMC13454116.txt.

Description

Zhouyang Xu, Hongwei Li, Wenchao Liu, and Haifeng Li present ScaleSpecter, a framework for reading brain wave recordings to sort neurodegenerative disease. EEG signals suit early detection, yet faint, fleeting faults hide inside drifting background rhythms. Their 2026 framework reads the signal at several time scales at once, from brief blips to long waves.

A light attention step lets the fine and coarse views inform each other. A spectral module then reweights the frequency parts, using both wave size and timing, to steer the reading. Tests ran on three public datasets covering Alzheimer disease, frontotemporal dementia, and Parkinson disease. The framework scored at or near the top on most measures, and cutaway tests showed each part adds value. The authors pitch it as a sturdy, general reader for messy brain signals.

Catalog record

CREATORZhouyang Xu, Hongwei Li, Wenchao Liu, Haifeng Li
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE77 KB
LICENSEhttps://creativecommons.org/licenses/by/4.0
RIGHTSCC BY
SOURCEEurope PMC

Source: https://europepmc.org/article/MED/PMC13454116

Browse the full library · About the Tower of Babel