Skip to content
Last Dictionary
Plane
Society, Ethics & EconomyMachine LearningMathematics & Statistics

individual fairness

nounAlso: fairness through awareness, Lipschitz fairness

In plain wordsThe idea that a fair system should treat similar people similarly: two applicants who are alike in every way that matters should get similar outcomes.

  1. A fairness criterion requiring that individuals who are similar with respect to the task receive similar outcomes, or similar distributions over outcomes. Formally, a randomized classifier M is individually fair if it satisfies a Lipschitz condition: the distance D between its output distributions for any two individuals is at most a task-specific similarity distance d between them. Unlike group fairnessgroup fairness, n.: Measuring fairness by comparing how a system treats whole groups, such as women and men, on average: their approval rates, their error rates or what their scores mean. criteria, it protects each person rather than group averages and cannot be met by treating some members of a group arbitrarily while the averages balance. Its central difficulty is the metric: deciding who counts as similar is itself a contested moral judgment, so later work tries to learn metrics from data or human judgments.

    The two applicants differed only in postcode, yet their scores were far apart, a clear violation of individual fairness.

D(M(x),M(x′))≤d(x,x′)for all individuals x,x′D\big(M(x), M(x')\big) \le d(x, x') \quad \text{for all individuals } x, x'

Origin

Introduced by Cynthia Dwork, Moritz Hardt, Toniann Pitassi, Omer Reingold and Richard Zemel in 'Fairness Through Awareness' (arXiv 2011; ITCS 2012), whose title contrasts the approach with the naive strategy of ignoring protected attributes.

Related

See also
  • group fairnessgroup fairness, n.: Measuring fairness by comparing how a system treats whole groups, such as women and men, on average: their approval rates, their error rates or what their scores mean.
  • counterfactual fairnesscounterfactual fairness, n.: A fairness test that asks: had this person been of another race or sex, with everything that would have followed from that, would the system have decided the same way?
  • algorithmic fairnessalgorithmic fairness, n.: The effort to make automated decisions fair: to say in numbers what fair treatment means, measure whether a system achieves it, and correct the system if it does not.
  • fairness through unawarenessfairness through unawareness, n.: The simple idea of making a system fair by hiding sensitive traits like race or sex from it; it usually fails, because other information gives those traits away.
  • demographic paritydemographic parity, n.: A fairness test that asks whether a system says yes (approves the loan, shortlists the applicant) at the same rate for every group, whatever other differences exist between them.
Contrast with
  • group fairnessgroup fairness, n.: Measuring fairness by comparing how a system treats whole groups, such as women and men, on average: their approval rates, their error rates or what their scores mean.

Sources

  1. Dwork, C., Hardt, M., Pitassi, T., Reingold, O. & Zemel, R. (2012). Fairness Through Awareness. Proc. 3rd Innovations in Theoretical Computer Science Conference (ITCS), 214–226. arxiv.org/abs/1104.3913

Mentioned in

algorithmic fairnessalgorithmic fairness, n.: The effort to make automated decisions fair: to say in numbers what fair treatment means, measure whether a system achieves it, and correct the system if it does not., counterfactual fairnesscounterfactual fairness, n.: A fairness test that asks: had this person been of another race or sex, with everything that would have followed from that, would the system have decided the same way?, Cynthia DworkCynthia Dwork, prop. n.: An American computer scientist who co-invented differential privacy, the leading mathematical definition of privacy in data analysis, and helped found the study of fairness in algorithms., demographic paritydemographic parity, n.: A fairness test that asks whether a system says yes (approves the loan, shortlists the applicant) at the same rate for every group, whatever other differences exist between them., fairness through unawarenessfairness through unawareness, n.: The simple idea of making a system fair by hiding sensitive traits like race or sex from it; it usually fails, because other information gives those traits away., group fairnessgroup fairness, n.: Measuring fairness by comparing how a system treats whole groups, such as women and men, on average: their approval rates, their error rates or what their scores mean.