GUESS: Generative Uncertainty Ensemble for Self Supervision

Salman Mohamadi, Gianfranco Doretto, Donald Adjeroh · 2024

Read GUESS: Generative Uncertainty Ensemble for Self Supervision by Salman Mohamadi, Gianfranco Doretto, Donald Adjeroh (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-2412-02896.txt.

Description

Salman Mohamadi, Gianfranco Doretto, and Donald Adjeroh present a self-supervised learning method in a 2024 paper. Self-supervised methods learn useful features from unlabeled images by solving a made-up task, then forcing the features to stay stable under changes to the image. The authors argue that forcing the same stability for every change can waste effort or even hurt later tasks.

Their method, GUESS, builds doubt about that stability into both the network design and its training loss. Slightly altered copies of each image pass through a group of networks, and a new loss blends making and sorting goals. A whitening-like step, softened by controlled noise, keeps the features spread out. Tests and part-by-part checks set GUESS as a strong new baseline for this style of learning.

Catalog record

CREATORSalman Mohamadi, Gianfranco Doretto, Donald Adjeroh
YEAR2024
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2412-02896

Browse the full library · About the Tower of Babel