What's in a Name? Reducing Bias in Bios without Access to Protected Attributes

被引:0
|
作者
Romanov, Alexey [1 ]
De-Arteaga, Maria [2 ]
Wallach, Hanna [3 ]
Chayes, Jennifer [3 ]
Borgs, Christian [3 ]
Chouldechova, Alexandra [2 ]
Geyik, Sahin [4 ]
Kenthapadi, Krishnaram [4 ]
Rumshisky, Anna [1 ]
Kalai, Adam Tauman [3 ]
机构
[1] Univ Massachusetts Lowell, Lowell, MA 01854 USA
[2] Carnegie Mellon Univ, Pittsburgh, PA USA
[3] Microsoft Res, New York, NY USA
[4] LinkedIn, Sunnyvale, CA USA
关键词
D O I
暂无
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
There is a growing body of work that proposes methods for mitigating bias in machine learning systems. These methods typically rely on access to protected attributes such as race, gender, or age. However, this raises two significant challenges: (1) protected attributes may not be available or it may not be legal to use them, and (2) it is often desirable to simultaneously consider multiple protected attributes, as well as their intersections. In the context of mitigating bias in occupation classification, we propose a method for discouraging correlation between the predicted probability of an individual's true occupation and a word embedding of their name. This method leverages the societal biases that are encoded in word embeddings, eliminating the need for access to protected attributes. Crucially, it only requires access to individuals' names at training time and not at deployment time. We evaluate two variations of our proposed method using a large-scale dataset of online biographies. We find that both variations simultaneously reduce race and gender biases, with almost no reduction in the classifier's overall true positive rate.
引用
收藏
页码:4187 / 4195
页数:9
相关论文
共 19 条
  • [1] What's in a Name? First Names as Facial Attributes
    Chen, Huizhong
    Gallagher, Andrew C.
    Girod, Bernd
    2013 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2013, : 3366 - 3373
  • [2] What's in a name? Children, young people and adultification bias
    Jenner, Rachel
    Lowey, Helen
    Dallimore, Damian
    ARCHIVES OF DISEASE IN CHILDHOOD, 2025,
  • [3] What’s in a Name: Exposing Gender Bias in Student Ratings of Teaching
    MacNell L.
    Driscoll A.
    Hunt A.N.
    Innovative Higher Education, 2015, 40 (4) : 291 - 303
  • [4] What's in a name? Frontotemporal dementia without cognitive impairment
    Lee, J.
    Finucane, T.
    JOURNAL OF THE AMERICAN GERIATRICS SOCIETY, 2014, 62 : S31 - S31
  • [5] What's in a Name? Can Name-Blind Evaluation Reduce Bias in AP Course Recommendation?
    Francis, Dania V.
    de Oliveira, Angela C. M.
    Dimmitt, Carey
    AEA PAPERS AND PROCEEDINGS, 2024, 114 : 205 - 209
  • [6] What’s in a name? Examining the effect of phonetic fit between spokesperson name and product attributes on source credibility
    Stacey Baxter
    Jasmina Ilicic
    Alicia Kulczynski
    Marketing Letters, 2015, 26 : 525 - 534
  • [7] What's in a name? Examining the effect of phonetic fit between spokesperson name and product attributes on source credibility
    Baxter, Stacey
    Ilicic, Jasmina
    Kulczynski, Alicia
    MARKETING LETTERS, 2015, 26 (04) : 525 - 534
  • [8] What's in a name? Measuring access to social activities with a field experiment
    Nesseler, Cornel
    Gomez-Gonzalez, Carlos
    Dietl, Helmut
    PALGRAVE COMMUNICATIONS, 2019, 5 (1)
  • [9] What’s in a name? Measuring access to social activities with a field experiment
    Cornel Nesseler
    Carlos Gomez-Gonzalez
    Helmut Dietl
    Palgrave Communications, 5
  • [10] What's in a Name? The Effects of the Labels "Fat" Versus "Overweight" on Weight Bias
    Brochu, Paula M.
    Esses, Victoria M.
    JOURNAL OF APPLIED SOCIAL PSYCHOLOGY, 2011, 41 (08) : 1981 - 2008