Refining the Information Bottleneck via Adversarial Information Separation

Shuai Ning, Zhenpeng Wang, Lin Wang, Bing Chen, Shuangrong Liu, Xu Wu, Jin Zhou, Bo Yang · 2026

Read Refining the Information Bottleneck via Adversarial Information Separation by Shuai Ning, Zhenpeng Wang, Lin Wang, Bing Chen, Shuangrong Liu, Xu Wu, Jin Zhou, Bo Yang (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-2602-06549.txt.

Description

Learning from scarce data is hardest when useful signals hide inside measurement noise, a daily problem in materials science. Shuai Ning and 7 co-authors present AdverISF, a framework that separates task-relevant features from noise without needing labels that say which is which. Standard tricks cannot draw that line precisely, and older rival methods demand explicit split labels.

AdverISF uses a self-supervised rival trick to force freedom between the useful features and the noise picture. A multi-layer design then recycles noise info across levels to recover features wrongly discarded as noise. Tests showed the method beat leading approaches on scarce data, and on real design tasks it fared better than rivals.

Catalog record

CREATORShuai Ning, Zhenpeng Wang, Lin Wang, Bing Chen, Shuangrong Liu, Xu Wu, Jin Zhou, Bo Yang
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE60 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-06549

Browse the full library · About the Tower of Babel