Fair Fairness Benchmark (FFB)
收藏arXiv2023-06-16 更新2024-06-21 收录
下载链接：
https://github.com/ahxt/fair_fairness_benchmark
下载链接
链接失效反馈官方服务：
资源简介：
Fair Fairness Benchmark (FFB) 是由德州农工大学和Meta AI等机构共同创建的一个标准化基准数据集，用于评估机器学习中的群体公平性方法。该数据集包含14个子数据集，涵盖了多种类型的数据，如表格数据和图像数据，旨在解决机器学习模型在不同群体中的公平性问题。FFB数据集通过提供统一的评估接口和详细的实验分析，帮助研究人员和实践者更好地理解和改进算法在不同群体间的公平性表现。
The Fair Fairness Benchmark (FFB) is a standardized benchmark dataset jointly created by Texas A&M University, Meta AI, and other institutions. It is developed to evaluate group fairness methodologies in machine learning. The dataset includes 14 sub-datasets that cover a wide range of data modalities such as tabular data and image data, with the objective of addressing fairness-related problems of machine learning models across different demographic groups. By providing a unified evaluation interface and detailed experimental analyses, the FFB dataset assists researchers and practitioners in better understanding and enhancing the fairness performance of algorithms across diverse groups.
提供机构：
德州农工大学
创建时间：
2023-06-16
