Three naive Bayes approaches for discrimination-free classification
By Toon Calders, Sicco Verwer. 2010
931 citations recorded by the source corpus.
In this paper, we investigate how to modify the naive Bayes classifier in order to perform classification that is restricted to be independent with respect to a given sensitive attribute. Such independency restrictions occur naturally when the decision process leading to the labels in the data-set was biased; e.g., due to gender or racial discrimination. This setting is motivated by many cases in which there exist laws that disallow a decision that is partly based on discrimination. Naive application of machine learning techniques would result in huge fines for companies. We present three approaches for making the naive Bayes classifier discrimination-free: (i) modifying the probability of the decision being positive, (ii) training one model for every sensitive attribute value and balancing them, and (iii) adding a latent variable to the Bayesian model that represents the unbiased label and optimizing the model parameters for likelihood using expectation maximization. We present experiments for the three approaches on both artificial and real-life data.
Subfields: Artificial Intelligence, Information Systems, AI Ethics, Fairness & Society
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Artificial Intelligence
- Silvia Chiappa
- Percy Liang
- Indrė Žliobaitė
- Luca Oneto
- Xintao Wu
- Mohit Bansal
- Sergey Levine
- Yoshua Bengio
In Information Systems
- Faisal Kamiran
- Sara Hajian
- Bettina Berendt
- Sören Preibusch
- Indrė Žliobaitė
- Josep Domingo-Ferrer
- Bryce Goodman
- Seth R. Flaxman
In AI Ethics, Fairness & Society
Citations in the corpus
- Towards fairer datasets: filtering and balancing the distribution of the people subtree in the ImageNet hierarchy (2019, background): “Researchers have proposed multiple fairness metrics including statistical parity [12, 13, 24, 39], disparate impact [27, 79], equalized odds [34] and individual fairness [23], and analyzed the relationship”
- Mitigating Gender Bias in Natural Language Processing: Literature Review (2019, background): “…bias have also been discussed extensively in artificial intelligence, machine learning, data mining, and several other application fields (e.g., (Calders and Verwer, 2010; Feldman et al., 2015; Hardt et al., 2016; Misra et al., 2016; Kleinberg et al., 2016; Pleiss et al., 2017; Beutel et al.,…”
- Inherent Trade-Offs in the Fair Determination of Risk Scores (2016, background): “Work in these directions has developed learning algorithms that penalize violations of statistical parity [4, 22].”
- Prediction-Based Decisions and Fairness: A Catalogue of Choices, Assumptions, and Definitions (2018, methodology): “[65, 11, 36, 61] give algorithms to build models achieving this notion of fairness”
- Conscientious Classification: A Data Scientist's Guide to Discrimination-Aware Classification (2017, neutral): “10,11 Post-processing is the final class of methods and can be performed post-training.”
- Iterative Orthogonal Feature Projection for Diagnosing Bias in Black-Box Models (2016, methodology): “In (Calders & Verwer, 2010) Kamiran & Calders present a method to transform data labels in order to remove discrimination.”
- Fair Prediction with Disparate Impact: A Study of Bias in Recidivism Prediction Instruments (2016, background): “While our discussion focusses primarily on first three fairness criteria, statistical parity is widely used within the machine learning community and may be the criterion with which many readers are most familiar[16, 17].”
- European Union Regulations on Algorithmic Decision-Making and a "Right to Explanation" (2016, methodology): “As Calders and Verwer note, ”postal code can reveal racial information and yet at the same time, still give useful, non-discriminatory information on loan defaulting” [9].”
- Data preprocessing techniques for classification without discrimination (2011, background): “In Discrimination-Aware Data Mining [21,20,22,16,5], two main research directions can be distinguished: the detection of discrimination [21,20,22], and the direction followed in this paper, namely learning classifiers if the data is discriminatory [16,5].”
- Causal Modeling-Based Discrimination Discovery and Removal: Criteria, Bounds, and Algorithms (2019, background): “sion tree [18], naive Bayes classifier [19], and logistic”
- Learning Optimal and Fair Decision Trees for Non-Discriminative Decision-Making (2019, background): “It consists in adding a fairness regularizer to the loss function objective, which serves to penalize discrimination, mitigating disparate treatment (Dwork et al. 2012; Zemel et al. 2013; Berk et al. 2017) or disparate impact (Calders and Verwer 2010; Kamiran, Calders, and Pech-enizkiy 2010).”
- Fairness Under Unawareness: Assessing Disparity When Protected Class Is Unobserved (2018, background): “The demographic disparity, or Calders-Verwer gap [5, 22], δ , is the difference in mean group outcomes between the advantaged and disadvantaged groups:”