BOOST: Out-of-Distribution-Informed Adaptive Sampling for Bias Mitigation in Stylistic Convolutional Neural Networks

2025-07-08

AI Summary

In art style classification, long-tailed training data distributions induce model bias—particularly degrading accuracy for rare styles. To address this, we propose BOOST: an out-of-distribution (OOD)-aware adaptive reweighting sampling framework that explicitly enhances learning weights for minority classes via dynamic temperature scaling and class-level sampling probability adjustment. BOOST is the first method to jointly model OOD detection and inter-class fairness optimization, introducing the fine-grained Style-wise Outlier Detection and Calibration (SODC) metric to quantify bias mitigation per class. Evaluated on KaoKore and PACS, BOOST significantly reduces inter-class bias (average reduction of 32.7%) while preserving high classification accuracy (decrease <0.8%), thereby achieving a balanced improvement in both fairness and accuracy.

Technology Category

• Computer Vision: Bias, Fairness & Privacy
• Machine Learning: Ethics, Bias, and Fairness
• Natural Language Processing: Ethics — Bias, Fairness, Transparency & Privacy

Application Category

• Social Networks and Social Media: Fairness and bias in social network and social media analysis
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking

Abstract

The pervasive issue of bias in AI presents a significant challenge to painting classification, and is getting more serious as these systems become increasingly integrated into tasks like art curation and restoration. Biases, often arising from imbalanced datasets where certain artistic styles dominate, compromise the fairness and accuracy of model predictions, i.e., classifiers are less accurate on rarely seen paintings. While prior research has made strides in improving classification performance, it has largely overlooked the critical need to address these underlying biases, that is, when dealing with out-of-distribution (OOD) data. Our insight highlights the necessity of a more robust approach to bias mitigation in AI models for art classification on biased training data. We propose a novel OOD-informed model bias adaptive sampling method called BOOST (Bias-Oriented OOD Sampling and Tuning). It addresses these challenges by dynamically adjusting temperature scaling and sampling probabilities, thereby promoting a more equitable representation of all classes. We evaluate our proposed approach to the KaoKore and PACS datasets, focusing on the model's ability to reduce class-wise bias. We further propose a new metric, Same-Dataset OOD Detection Score (SODC), designed to assess class-wise separation and per-class bias reduction. Our method demonstrates the ability to balance high performance with fairness, making it a robust solution for unbiasing AI models in the art domain.

Problem

• Mitigating bias in AI for painting classification
• Addressing imbalanced datasets in artistic style recognition
• Improving fairness in model predictions for OOD data

Innovation

• OOD-informed adaptive sampling for bias mitigation
• Dynamic temperature scaling adjustment for fairness
• Same-Dataset OOD Detection Score (SODC) metric