Fighting Selection Bias in Statistical Learning: Application to Visual Recognition from Biased Image Databases

TL;DR

This paper introduces a method to reduce selection bias in visual recognition systems by reweighting biased datasets using shared low-dimensional image representations, improving fairness and accuracy.

Contribution

It proposes a bias correction technique based on reweighting observations with known bias mechanisms, applicable to deep neural network training on biased image datasets.

Findings

Reweighting improves recognition fairness across population segments.

Shared low-dimensional representations enable effective bias correction.

Numerical experiments demonstrate the approach's effectiveness.

Abstract

In practice, and especially when training deep neural networks, visual recognition rules are often learned based on various sources of information. On the other hand, the recent deployment of facial recognition systems with uneven performances on different population segments has highlighted the representativeness issues induced by a naive aggregation of the datasets. In this paper, we show how biasing models can remedy these problems. Based on the (approximate) knowledge of the biasing mechanisms at work, our approach consists in reweighting the observations, so as to form a nearly debiased estimator of the target distribution. One key condition is that the supports of the biased distributions must partly overlap, and cover the support of the target distribution. In order to meet this requirement in practice, we propose to use a low dimensional image representation, shared across the image databases. Finally, we provide numerical experiments highlighting the relevance of our approach.