TY - GEN
T1 - CFP
T2 - 33rd International Conference on Artificial Neural Networks, ICANN 2024
AU - Zhang, Simiao
AU - Bai, Jitao
AU - Guan, Menghong
AU - Zhang, Yueling
AU - Sun, Jun
AU - Huang, Yihao
AU - Wang, Jiaping
AU - Wan, Cheng Cheng
AU - Su, Ting
AU - Pu, Geguang
N1 - Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2024.
PY - 2024
Y1 - 2024
N2 - Machine learning models are increasingly used for impactful decisions, such as loan approval, criminal sentencing, and resume filtering, raising concerns about ensuring fairness without sacrificing performance. However, fairness has multiple definitions, and existing techniques targeting specific metrics have limitations in improving multiple notions of fairness simultaneously. In this work, we establish a comprehensive measurement to simultaneously consider multiple fairness notions as well as performance, and propose new metrics through an in-depth analysis of the relationship between different fairness metrics. Based on the comprehensive measurement and new metrics, we present CFP, a reinforcement learning-based framework, to efficiently improve the fairness-performance trade-off in machine learning classifiers. We conduct extensive experiments to evaluate CFP on 6 tasks, 3 machine learning models, and 15 fairness-performance measurements. The results demonstrate that CFP can improve the classifiers on multiple fairness metrics without sacrificing its performance.
AB - Machine learning models are increasingly used for impactful decisions, such as loan approval, criminal sentencing, and resume filtering, raising concerns about ensuring fairness without sacrificing performance. However, fairness has multiple definitions, and existing techniques targeting specific metrics have limitations in improving multiple notions of fairness simultaneously. In this work, we establish a comprehensive measurement to simultaneously consider multiple fairness notions as well as performance, and propose new metrics through an in-depth analysis of the relationship between different fairness metrics. Based on the comprehensive measurement and new metrics, we present CFP, a reinforcement learning-based framework, to efficiently improve the fairness-performance trade-off in machine learning classifiers. We conduct extensive experiments to evaluate CFP on 6 tasks, 3 machine learning models, and 15 fairness-performance measurements. The results demonstrate that CFP can improve the classifiers on multiple fairness metrics without sacrificing its performance.
KW - Ethics of AI
KW - Fairness in machine learning
KW - Fairness-performance trade-off
KW - Reinforcement learning
UR - https://www.scopus.com/pages/publications/85205101392
U2 - 10.1007/978-3-031-72332-2_30
DO - 10.1007/978-3-031-72332-2_30
M3 - 会议稿件
AN - SCOPUS:85205101392
SN - 9783031723315
T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
SP - 463
EP - 477
BT - Artificial Neural Networks and Machine Learning – ICANN 2024 - 33rd International Conference on Artificial Neural Networks, Proceedings
A2 - Wand, Michael
A2 - Schmidhuber, Jürgen
A2 - Wand, Michael
A2 - Malinovská, Kristína
A2 - Schmidhuber, Jürgen
A2 - Tetko, Igor V.
A2 - Tetko, Igor V.
PB - Springer Science and Business Media Deutschland GmbH
Y2 - 17 September 2024 through 20 September 2024
ER -
