Null-sampling for interpretable and fair representations

Kehrenberg, Thomas, Bartlett, Myles, Thomas, Oliver and Quadrianto, Novi (2020) Null-sampling for interpretable and fair representations. European Conference on Computer Vision (ECCV), Online, 23 Aug 2020 - 28 Aug 2020. Published in: Computer Vision – ECCV 2020. 12373 Springer International Publishing ISBN 9783030586041

[img] PDF - Accepted Version
Download (3MB)


We propose to learn invariant representations, in the data domain, to achieve interpretability in algorithmic fairness. Invariance implies a selectivity for high level, relevant correlations w.r.t. class label annotations, and a robustness to irrelevant correlations with protected characteristics such as race or gender. We introduce a non-trivial setup in which the training set exhibits a strong bias such that class label annotations are irrelevant and spurious correlations cannot be distinguished. To address this problem, we introduce an adversarially trained model with a null-sampling procedure to produce invariant representations in the data domain. To enable disentanglement, a partially-labelled representative set is used. By placing the representations into the data domain, the changes made by the model are easily examinable by human auditors. We show the effectiveness of our method on both image and tabular datasets: Coloured MNIST, the CelebA and the Adult dataset.

Item Type: Conference Proceedings
Keywords: algorithmic fairness, interpretability in fairness, adversarial learning, invertible neural network, variational autoencoder
Schools and Departments: School of Engineering and Informatics > Informatics
SWORD Depositor: Mx Elements Account
Depositing User: Mx Elements Account
Date Deposited: 18 Aug 2020 07:08
Last Modified: 23 Feb 2022 12:10

View download statistics for this item

📧 Request an update
Project NameSussex Project NumberFunderFunder Ref
BayesianGDPR - Bayesian Models and Algorithms for Fairness and TransparencyG2903EUROPEAN UNIONUnset