BOOST: Out-of-Distribution-Informed Adaptive Sampling for Bias Mitigation in Stylistic Convolutional Neural Networks Journal: Expert Systems With Applications

Mridula Vijendran, Shuang Chen, Jingjing Deng, Hubert P. H. Shum · 2025

Read BOOST: Out-of-Distribution-Informed Adaptive Sampling for Bias Mitigation in Stylistic Convolutional Neural Networks Journal: Expert Systems With Applications by Mridula Vijendran, Shuang Chen, Jingjing Deng, Hubert P. H. Shum (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-2507-07134.txt.

Description

Mridula Vijendran, Shuang Chen, Jingjing Deng, and Hubert Shum present BOOST, a sampling method that fights bias in AI models that classify paintings by style, in a 2025 paper. Training sets often overflow with a few styles, so models learn the rare styles poorly. Their method watches for classes the model finds strange and adjusts how often each class is sampled during training.

It also tunes a confidence setting per class and comes with a new score for measuring class-wise bias on familiar data. On two art data sets, the method balanced accuracy across styles better than rivals. The authors aim at fairer tools for curation and restoration, where skewed calls carry real costs.

Catalog record

CREATORMridula Vijendran, Shuang Chen, Jingjing Deng, Hubert P. H. Shum
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE63 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2507-07134

Browse the full library · About the Tower of Babel