Lee Joshua, Bu Yuheng, Sattigeri Prasanna, Panda Rameswar, Wornell Gregory W, Karlinsky Leonid, Schmidt Feris Rogerio
Department of Electrical Engineering and Computer Science, Massachusetts Institute of Technology, Cambridge, MA 02139, USA.
MIT-IBM Watson AI Lab, IBM Research, Cambridge, MA 02139, USA.
Entropy (Basel). 2022 Mar 26;24(4):461. doi: 10.3390/e24040461.
As machine learning algorithms grow in popularity and diversify to many industries, ethical and legal concerns regarding their fairness have become increasingly relevant. We explore the problem of algorithmic fairness, taking an information-theoretic view. The maximal correlation framework is introduced for expressing fairness constraints and is shown to be capable of being used to derive regularizers that enforce independence and separation-based fairness criteria, which admit optimization algorithms for both discrete and continuous variables that are more computationally efficient than existing algorithms. We show that these algorithms provide smooth performance-fairness tradeoff curves and perform competitively with state-of-the-art methods on both discrete datasets (COMPAS, Adult) and continuous datasets (Communities and Crimes).
随着机器学习算法在众多行业中日益普及并实现多样化，关于其公平性的伦理和法律问题变得越来越重要。我们从信息论的角度探讨算法公平性问题。引入了最大相关框架来表达公平性约束，并证明该框架能够用于推导强制执行基于独立性和分离性的公平性标准的正则化器，这些正则化器允许针对离散和连续变量的优化算法，其计算效率高于现有算法。我们表明，这些算法提供了平滑的性能-公平性权衡曲线，并且在离散数据集（COMPAS、成人）和连续数据集（社区与犯罪）上与最先进的方法相比具有竞争力。
