Fairness Through Causal Awareness: Learning Latent-Variable Models for Biased Data
arXiv:1809.02519
Abstract
How do we learn from biased data? Historical datasets often reflect historical prejudices; sensitive or protected attributes may affect the observed treatments and outcomes. Classification algorithms tasked with predicting outcomes accurately from these datasets tend to replicate these biases. We advocate a causal modeling approach to learning from biased data, exploring the relationship between fair classification and intervention. We propose a causal model in which the sensitive attribute confounds both the treatment and the outcome. Building on prior work in deep learning and generative modeling, we describe how to learn the parameters of this causal model from observational data alone, even in the presence of unobserved confounders. We show experimentally that fairness-aware causal modeling provides better estimates of the causal effects between the sensitive attribute, the treatment, and the outcome. We further present evidence that estimating these causal effects can help learn policies that are both more accurate and fair, when presented with a historically biased dataset.
arXiv commentAccepted as a conference paper at ACM Conference on Fairness, Accountability, and Transparency (ACM FAT*) 2019
References in corpus (12)
- Auto-Encoding Variational Bayes
- Stochastic Backpropagation and Approximate Inference in Deep Generative Models
- Equality of Opportunity in Supervised Learning
- Counterfactual Fairness
- On the (im)possibility of fairness
- Fairness Without Demographics in Repeated Loss Minimization
- Learning Representations for Counterfactual Inference
- Towards a Neural Statistician
- Residual Unfairness in Fair Machine Learning from Prejudiced Data
- Interventions over Predictions: Reframing the Ethical Debate for Actuarial Risk Assessment
- Causal Interventions for Fairness
- Model Criticism for Bayesian Causal Inference