节点嵌入中通过受控(不)感知实现的公平性
Fairness Through Controlled (Un)Awareness in Node Embeddings
AI总结:
本文研究图表示学习中的公平性问题,通过微调CrossWalk算法的超参数来控制节点嵌入中敏感属性的可检测性,从而提升机器学习系统对不同公平性范式的适应性。
摘要:
图表示学习对于将机器学习(ML)模型应用于复杂图(如社交网络)至关重要。由于社会影响和敏感个人数据的使用,确保“公平”的表示是必不可少的。在本文中,我们展示了CrossWalk算法的参数化如何影响从节点嵌入中推断敏感属性的能力。通过微调超参数,我们表明可以显著增强或掩盖这些属性的可检测性。此功能为改善利用图嵌入的ML系统的公平性提供了有价值的工具,使其适应不同的公平性范式。
英文摘要:
Graph representation learning is central for the application of machine learning (ML) models to complex graphs, such as social networks. Ensuring `fair' representations is essential, due to the societal implications and the use of sensitive personal data. In this paper, we demonstrate how the parametrization of the \emph{CrossWalk} algorithm influences the ability to infer a sensitive attributes from node embeddings. By fine-tuning hyperparameters, we show that it is possible to either significantly enhance or obscure the detectability of these attributes. This functionality offers a valuable tool for improving the fairness of ML systems utilizing graph embeddings, making them adaptable to different fairness paradigms.
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。