arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2203.11852cs.DBcs.LG
收藏

数据中的表示偏差:识别与消解技术综述

Representation Bias in Data: A Survey on Identification and Resolution Techniques

University of Illinois Chicago(伊利诺伊大学芝加哥分校);University of Michigan(密歇根大学)

Nima Shahbazi, Yin Lin, Abolfazl Asudeh, H. V. Jagadish

摘要:

数据驱动的算法只与它们所使用的数据一样好,而数据集,特别是社交数据,往往不能充分代表少数群体。数据中的代表性偏差可能由于各种原因而发生,从历史歧视到数据采集和准备方法中的选择和抽样偏差。鉴于“偏见,偏见”,人们不能指望基于人工智能的解决方案在不解决代表性偏见等问题的情况下为社会应用带来公平的结果。虽然对机器学习模型中的公平性进行了广泛的研究,包括几篇综述论文,但对数据中的偏见研究较少。本文回顾了文献识别和解决代表性偏差作为一个数据集的功能,独立于如何消费以后。本调查的范围限于结构化(表格)和非结构化(例如,图像、文本、图形)数据。它提出了基于多个设计维度对所研究的技术进行分类的分类法,并对它们的属性进行了并排比较。要完全解决数据中的代表性偏见问题,还有很长的路要走。作者希望这项调查能够激励研究人员通过观察各自领域内的现有工作来应对未来的挑战。

英文摘要:

Data-driven algorithms are only as good as the data they work with, while data sets, especially social data, often fail to represent minorities adequately. Representation Bias in data can happen due to various reasons ranging from historical discrimination to selection and sampling biases in the data acquisition and preparation methods. Given that "bias in, bias out", one cannot expect AI-based solutions to have equitable outcomes for societal applications, without addressing issues such as representation bias. While there has been extensive study of fairness in machine learning models, including several review papers, bias in the data has been less studied. This paper reviews the literature on identifying and resolving representation bias as a feature of a data set, independent of how consumed later. The scope of this survey is bounded to structured (tabular) and unstructured (e.g., image, text, graph) data. It presents taxonomies to categorize the studied techniques based on multiple design dimensions and provides a side-by-side comparison of their properties. There is still a long way to fully address representation bias issues in data. The authors hope that this survey motivates researchers to approach these challenges in the future by observing existing work within their respective domains.

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。

↑