首页/论文/1901.09565

arXiv 1901.095652019-01-28cs.LGstat.ML

Fairness in representation: quantifying stereotyping as a representational harm

University of Utah（犹他大学）；Haverford College（哈弗福德学院）；University of Arizona（亚利桑那大学）

Mohsen Abbasi, Sorelle A. Friedler, Carlos Scheidegger, Suresh Venkatasubramanian

备注：: 9 pages, 6 figures, Siam International Conference on Data Mining

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

英文摘要相关论文

英文摘要：

While harms of allocation have been increasingly studied as part of the subfield of algorithmic fairness, harms of representation have received considerably less attention. In this paper, we formalize two notions of stereotyping and show how they manifest in later allocative harms within the machine learning pipeline. We also propose mitigation strategies and demonstrate their effectiveness on synthetic datasets.

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。

阅读原文

阅读 PDF ↗

arXiv 页面 ↗HTML 原文 ↗

本文内容

英文摘要相关论文

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。

继续探索

相关论文

Towards Understanding and Mitigating Social Biases in Language Models

Carnegie Mellon University（卡内基梅隆大学）

As machine learning methods are deployed in real-world settings such as healthcare, legal systems, and social science, it is crucial to recognize how they shape social biases and stereotypes in these sensitive decision-making processes. Among such real-world deployments are large-scale pretrained language models (LMs) that can be potentially dangerous in manifesting undesirable representational biases - harmful biases resulting from stereotyping that propagate negative generalizations involving gender, race, religion, and other social constructs. As a step towards improving the fairness of LMs…

2021-06-24cs.AIcs.CLcs.CY

FairDistillation: Mitigating Stereotyping in Language Models

KU Leuven（鲁汶大学）；Leuven.AI institute（鲁汶人工智能研究所）；TU Berlin（柏林工业大学）；Weizenbaum Institute（魏泽鲍姆研究所）

Large pre-trained language models are successfully being used in a variety of tasks, across many languages. With this ever-increasing usage, the risk of harmful side effects also rises, for example by reproducing and reinforcing stereotypes. However, detecting and mitigating these harms is difficult to do in general and becomes computationally expensive when tackling multiple languages or when considering different biases. To address this, we present FairDistillation: a cross-lingual method based on knowledge distillation to construct smaller language models while controlling for specific bias…

2022-09-16cs.CLcs.CYcs.LG

超越行为主义表征性伤害：测量与缓解的计划

Beyond Behaviorist Representational Harms: A Plan for Measurement and Mitigation

UC San Diego（加利福尼亚大学圣迭戈分校）

本文通过审视现有表征性伤害定义的不足，提出超越行为主义、纳入认知与情感伤害的扩展框架，并结合大语言模型案例给出测量要求与缓解措施，以推动公平性研究走向实践。

2024-05-06cs.AIcs.CLcs.CY

从代表性伤害到服务质量伤害：以 Llama 2 安全防护措施为例的案例研究

From Representational Harms to Quality-of-Service Harms: A Case Study on Llama 2 Safety Safeguards

McGill University（麦吉尔大学）；University of Montreal（蒙特利尔大学）；Mila - Quebec AI Institute（米拉-魁北克人工智能研究所）

本文以 Llama 2 为例，通过构建无毒提示和新的响应分类体系，揭示大语言模型安全响应仍可能编码有害假设，并发现安全性与有用性的权衡对边缘化群体更明显，从而导致服务质量伤害。

2024-07-05cs.CLcs.CYcs.LG

对比学习中表征损害与分配损害的研究

An Investigation of Representation and Allocation Harms in Contrastive Learning

University of Michigan（密歇根大学）；IBM Research（IBM研究院）；MIT-IBM Watson AI lab（MIT-IBM沃森人工智能实验室）

本研究针对自监督学习中少数群体表征损害研究不足的问题，验证了对比学习会引发少数群体表征坍缩的现象，通过因果分析明确其与下游分配损害的关联，并给出理论解释。

2023-10-02cs.LGstat.ML

生成模型增强推荐系统相关的算法危害

Algorithmic Harms Associated with Generative Model-Augmented Recommendation Systems

本研究扩展了算法危害分类法，分析生成模型增强推荐系统可能引发的新型及内生危害，并提供因果分析以支持检测与缓解。

2026-09-29cs.AIcs.CLcs.IR

更多相关论文收起更多论文

Image Cropping on Twitter: Fairness Metrics, their Limitations, and the Importance of Representation, Design, and Agency

Twitter（推特）

Twitter uses machine learning to crop images, where crops are centered around the part predicted to be the most salient. In fall 2020, Twitter users raised concerns that the automated image cropping system on Twitter favored light-skinned over dark-skinned individuals, as well as concerns that the system favored cropping woman's bodies instead of their heads. In order to address these concerns, we conduct an extensive analysis using formalized group fairness metrics. We find systematic disparities in cropping and identify contributing factors, including the fact that the cropping based on the …

2021-09-09cs.CVcs.CYcs.HC

放任的伤害：生成语言模型中的算法偏见

Laissez-Faire Harms: Algorithmic Biases in Generative Language Models

Young Data Scientists League（年轻数据科学家联盟）；Stanford University（斯坦福大学）；Georgia Institute of Technology（佐治亚理工学院）；Schar School of Policy and Government & Department of Computer Science（政策与政府学院及计算机科学系）；George Mason University（乔治·马歇尔大学）

研究通过开放性提示分析生成语言模型的偏见，发现五种主流模型对少数族裔、性别和性取向群体存在排除、从属和刻板印象的伤害，强调需保护消费者免受语言模型歧视性影响。

2024-04-16cs.AIcs.CLcs.CY

学习类人表征以赋能人类价值观的学习

Learning Human-like Representations to Enable Learning Human Values

Johns Hopkins University（约翰斯·霍普金斯大学）；Princeton University（普林斯顿大学）；New York University（纽约大学）

本文研究通过人类与AI智能体间的表征对齐，使AI能安全快速地学习个体人类价值观，并在多臂老虎机实验中证明该方法支持安全探索并提升泛化能力。

2024-11-08cs.AIcs.LG

机器学习公平性中平等的意义：超越机会的平等

What Is the Point of Equality in Machine Learning Fairness? Beyond Equality of Opportunity

Seoul National University（首尔国立大学）

本文探讨机器学习公平性中平等的意义，提出超越机会平等的平等主义框架，以解决结构性不平等和代表性伤害问题。

2025-09-01cs.AIcs.CYcs.LG

基于Slepian函数的局部高分辨率地理表示

Localized, High-resolution Geographic Representations with Slepian Functions

Department of Computer Science, University of Colorado Boulder（科罗拉多大学波德分校计算机科学系）；Microsoft（微软公司）；University College London（伦敦大学学院）；McGill University（麦吉尔大学）；Mila–Quebec AI Institute（魁北克AI研究所）；University of Bonn, Germany（德国波恩大学）；University of Wageningen, Netherlands（荷兰瓦赫宁根大学）

提出利用球面Slepian函数构建地理编码器，在感兴趣区域内集中表示能力，实现高分辨率且计算高效，并引入混合Slepian-球谐编码器平衡局部与全局性能，在分类、回归等任务中优于基线。

2026-06-02cs.LG

趋同的世界表征与分歧的任务

Convergent World Representations and Divergent Tasks

Center for Brain Science, Harvard University（哈佛大学脑科学中心）；CBS-NTT Program in Physics of Intelligence, Harvard University（哈佛大学物理智能项目）；Prior Computers, Cambridge, MA（剑桥马萨诸塞州Prior公司）

本研究通过多任务训练产生趋同的世界表征，但存在分歧任务会损害新实体的整合与泛化。

2026-01-31cs.AIcs.LG

Joint Multisided Exposure Fairness for Recommendation

McGill University（麦吉尔大学）；Microsoft（微软）；City University of Hong Kong（香港城市大学）；Google（谷歌）

Prior research on exposure fairness in the context of recommender systems has focused mostly on disparities in the exposure of individual or groups of items to individual users of the system. The problem of how individual or groups of items may be systemically under or over exposed to groups of users, or even all users, has received relatively less attention. However, such systemic disparities in information exposure can result in observable social harms, such as withholding economic opportunities from historically marginalized groups (allocative harm) or amplifying gendered and racialized ste…

2022-04-29cs.AIcs.IRcs.LG