Gender Shades: Intersectional Accuracy Disparities in Commercial Gender Classification

نویسندگان

  • Joy Buolamwini
  • Timnit Gebru
چکیده

Recent studies demonstrate that machine learning algorithms can discriminate based on classes like race and gender. In this work, we present an approach to evaluate bias present in automated facial analysis algorithms and datasets with respect to phenotypic subgroups. Using the dermatologist approved Fitzpatrick Skin Type classification system, we characterize the gender and skin type distribution of two facial analysis benchmarks, IJB-A and Adience. We find that these datasets are overwhelmingly composed of lighter-skinned subjects (79.6% for IJB-A and 86.2% for Adience) and introduce a new facial analysis dataset which is balanced by gender and skin type. We evaluate 3 commercial gender classification systems using our dataset and show that darker-skinned females are the most misclassified group (with error rates of up to 34.7%). The maximum error rate for lighter-skinned males is 0.8%. The substantial disparities in the accuracy of classifying darker females, lighter females, darker males, and lighter males in gender classification systems require urgent attention if commercial companies are to build genuinely fair, transparent and accountable facial analysis algorithms.

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

A Comparative Study of Gender and Age Classification in Speech Signals

Accurate gender classification is useful in speech and speaker recognition as well as speech emotion classification, because a better performance has been reported when separate acoustic models are employed for males and females. Gender classification is also apparent in face recognition, video summarization, human-robot interaction, etc. Although gender classification is rather mature in a...

متن کامل

An intersectional approach to Men's Health

Men’s lives and health are rooted in opportunity structures that are shaped by race, ethnicity and other characteristics that have important social, political, economic and cultural meaning. Within men’s health, there is a need to consider how structural factors and men’s socially-defined characteristics affect the relationship between sex, gender and health. The goal of an intersectional appro...

متن کامل

Gender and Climate Change Adaptation in Agrarian Settings: Current Thinking, New Directions, and Research Frontiers

The impacts of climate variability and change impinge upon different lives and livelihoods within agrarian populations in complex ways. While academic, donor, and implementer efforts to understand and act on this complexity have been profoundly influenced by gender analysis, most contemporary analyses are predicated on a construction of gender as binary (men versus women). This approach runs co...

متن کامل

Self-Rated Health Trajectories among Married Americans: Do Disparities Persist over 20 Years?

The purpose of this study is to understand self-rated health (SRH) trajectories by social location (race/ethnicity by gender by social class) among married individuals in the United States. We estimate multilevel models of SRH using six observations from 1980 to 2000 from a nationally representative panel of married individuals initially aged 25-55 (Marital Instability Over the Life Course Stud...

متن کامل

(En)gendering racial disparities in health trajectories: A life course and intersectional analysis

Historically, intersectionality has been an underutilized framework in sociological research on racial/ethnic and gender inequalities in health. To demonstrate its utility and importance, we conduct an intersectional analysis of the social stratification of health using the exemplar of hypertension-a health condition in which racial/ethnic and gender differences have been well-documented. Previ...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2018