Flexibly Fair Representation Learning by Disentanglement

By Elliot Creager, David Madras, Jörn-Henrik Jacobsen, Marissa A. Weis, Kevin Swersky, Toniann Pitassi, Richard S. Zemel. 2019

383 citations recorded by the source corpus.

We consider the problem of learning representations that achieve group and subgroup fairness with respect to multiple sensitive attributes. Taking inspiration from the disentangled representation learning literature, we propose an algorithm for learning compact representations of datasets that are useful for reconstruction and prediction, but are also \emph{flexibly fair}, meaning they can be easily modified at test time to achieve subgroup demographic parity with respect to multiple sensitive attributes and their conjunctions. We show empirically that the resulting encoder---which does not require the sensitive attributes for inference---enables the adaptation of a single representation to a variety of fair classification tasks with new target labels and subgroup definitions.

Subfields: Artificial Intelligence, Safety Research, Representation & Self-Supervised Learning

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

In Safety Research

In Representation & Self-Supervised Learning

Citations in the corpus

Reducing Sentiment Bias in Language Models via Counterfactual Evaluation (2019, background): “…of inputs in the language models, contributing to a family of methods to learn fair representations (Beutel et al., 2017; Zemel et al., 2013; Creager et al., 2019; Edwards and Storkey, 2016; Louizos et al., 2016) and enforcing independence between sensitive attributes and prediction outputs…”
Shortcut learning in deep neural networks (2020, background): “class-predictive pixels or image quality) and record the accuracy drop on clean test data as a way of finding out how prone a given architecture and loss function are to picking up on shortcuts [39, 40, 100, 41].”
A Survey on Bias and Fairness in Machine Learning (2019, background): “From learning fair representations [38, 82, 92] to learning fair word embeddings [19, 50, 141], debiasing methods have been proposed in different AI applications and domains.”
Weakly-Supervised Disentanglement Without Compromises (2020, background): “This result is of great importance as the community is becoming increasingly interested in the practical benefits of disentangled representations (van Steenkiste et al., 2019; Locatello et al., 2019a; Creager et al., 2019; Chao et al., 2019; Iten et al., 2020; Chartsias et al., 2019; Higgins et al.,”
On the Transfer of Inductive Bias from Simulation to the Real World: a New Disentanglement Dataset (2019, background): “Despite the intended applications of disentangled representation learning algorithms to real data in ﬁelds such as robotics, healthcare and fair decision making [e.g. 6, 10, 20], state-of-the-art approaches have only been systematically evaluated on synthetic toy datasets.”
Towards Fairness in Visual Recognition: Effective Strategies for Bias Mitigation (2019, background): “Additionally, recent works have shown that ML models possess bias towards legally protected classes [29, 6, 4, 7, 33, 8].”
Towards Nonlinear Disentanglement in Natural Data with Temporal Sparse Coding (2020, background): “A disentangled representation would be valuable for a wide variety of machine learning applications, including sample efﬁciency for downstream tasks [8, 26], fairness [27, 28] and interpretability [1, 29, 30].”
Learning De-biased Representations with Biased Representations (2019, positive): “, 2019b) or mutual information minimisation (Kim et al., 2019; Creager et al., 2019).”
Guided Variational Autoencoder for Disentanglement Learning (2020, background): “Recent efforts in fairness disentanglement learning [9, 47] also bear some similarity, but there is still a large difference in formulation.”
Is Independence all you need? On the Generalization of Representations Learned from Correlated Data (2020, methodology): “Given the potential societal impact in the medical domain (Chartsias et al., 2018) or fair decision making (Lo-catello et al., 2019a; Madras et al., 2018; Creager et al., 2019), the evaluation of the usefulness of disentangled representations trained on correlated data is of critical importance.”
Benchmarking Bias Mitigation Algorithms in Representation Learning through Fairness Metrics (2021, methodology): “…initially within the framework of Generative Adversarial Networks (GANs) [35] – has also been extensively leveraged in the bias-mitigation literature [6, 7, 8, 9, 26, 27, 28, 29, 30, 31, 32, 33] to make diﬀerent groups indistinguishable from one another with respect to a sensitive attribute.”
Say No to the Discrimination: Learning Fair Graph Neural Networks with Limited Sensitive Attribute Information (2020, background): “Second, though extensive efforts have been made to establish fair models by revising features [22, 23, 56], disentanglement [9, 31], adversarial debiasing [3, 13, 33, 55] and fairness constraints [52, 53], they are overwhelmingly dedicated to independently and identically distributed (i.”
Disentangling Influence: Using Disentangled Representations to Audit Model Predictions (2019, background): “Disentanglement has recently been shown to be a useful property for learning and evaluating fair machine learning models [5, 17].”
Algorithm fairness in artificial intelligence for medicine and healthcare (2023, background): “…can be viewed as a form of data pre-processing for debiasing representations in downstream fairness tasks and for providing flexibility in terms of allowing data users to isolate and truncate specific latent codes that correspond to protected attributes in representation space 255,292,293 .”