Fairness-aware Class Imbalanced Learning

TL;DR

This paper evaluates long-tail learning methods for NLP tasks like sentiment and occupation classification, extending margin-loss approaches to improve fairness and reduce bias in imbalanced datasets.

Contribution

It introduces fairness-aware extensions to class imbalance techniques and empirically demonstrates their effectiveness in mitigating bias.

Findings

Proposed methods reduce demographic bias in NLP tasks.

Extended margin-loss approach improves fairness in long-tail distributions.

Controlled experiments confirm effectiveness in class imbalance scenarios.

Abstract

Class imbalance is a common challenge in many NLP tasks, and has clear connections to bias, in that bias in training data often leads to higher accuracy for majority groups at the expense of minority groups. However there has traditionally been a disconnect between research on class-imbalanced learning and mitigating bias, and only recently have the two been looked at through a common lens. In this work we evaluate long-tail learning methods for tweet sentiment and occupation classification, and extend a margin-loss based approach with methods to enforce fairness. We empirically show through controlled experiments that the proposed approaches help mitigate both class imbalance and demographic biases.