arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2303.08516cs.LGcs.CY
收藏

基于观测数据的公平离策略学习

Fair Off-Policy Learning from Observational Data

LMU Munich(慕尼黑大学);Munich Center for Machine Learning(慕尼黑机器学习中心)

Dennis Frauen, Valentyn Melnychuk, Stefan Feuerriegel

AI总结:

针对离策略学习公平性研究不足的问题,提出基于神经网络的公平离策略学习框架,形式化多种公平概念、给出有限样本泛化界,经实验验证有效,可支撑公平性要求下的算法决策应用。

摘要:

出于法律、伦理和社会层面的原因,实际应用中的算法决策必须具备公平性。为实现这一目标,已有研究提出了多种保障机器学习预测公平性的方法,但相比之下,针对决策过程公平性——尤其是离策略学习公平性的研究投入较少。本文提出了一种新颖的公平离策略学习框架:我们在不同公平性概念下从观测数据中学习决策规则,其中明确假设观测数据是在一种不同的、可能存在歧视的行为策略下收集的。为此,我们首先形式化定义了离策略学习的不同公平性概念,随后提出一种基于神经网络的框架,用于在不同公平性概念下学习最优策略。我们还以泛化界的形式,为该框架的有限样本版本提供了理论保证。我们通过使用模拟数据和真实世界数据开展的大量数值实验,验证了该框架的有效性。总体而言,我们的工作为众多需要保障公平性的实际应用场景中的算法决策提供了支持。

英文摘要:

Algorithmic decision-making in practice must be fair for legal, ethical, and societal reasons. To achieve this, prior research has contributed various approaches that ensure fairness in machine learning predictions, while comparatively little effort has focused on fairness in decision-making, specifically off-policy learning. In this paper, we propose a novel framework for fair off-policy learning: we learn decision rules from observational data under different notions of fairness, where we explicitly assume that observational data were collected under a different potentially discriminatory behavioral policy. For this, we first formalize different fairness notions for off-policy learning. We then propose a neural network-based framework to learn optimal policies under different fairness notions. We further provide theoretical guarantees in the form of generalization bounds for the finite-sample version of our framework. We demonstrate the effectiveness of our framework through extensive numerical experiments using both simulated and real-world data. Altogether, our work enables algorithmic decision-making in a wide array of practical applications where fairness must be ensured.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑