Women also Snowboard: Overcoming Bias in Captioning Models

TL;DR

This paper introduces the Equalizer model to reduce gender bias in image captioning, ensuring fairer gender predictions by focusing on visual evidence rather than contextual cues, and demonstrates improved accuracy and bias mitigation.

Contribution

The paper proposes the Equalizer model with novel loss functions to mitigate gender bias in captioning models, emphasizing visual evidence over contextual cues.

Findings

Lower gender prediction error compared to prior models.

More accurate gender representation matching ground truth.

Model more often looks at people when predicting gender.

Abstract

Most machine learning methods are known to capture and exploit biases of the training data. While some biases are beneficial for learning, others are harmful. Specifically, image captioning models tend to exaggerate biases present in training data (e.g., if a word is present in 60% of training sentences, it might be predicted in 70% of sentences at test time). This can lead to incorrect captions in domains where unbiased captions are desired, or required, due to over-reliance on the learned prior and image context. In this work we investigate generation of gender-specific caption words (e.g. man, woman) based on the person's appearance or the image context. We introduce a new Equalizer model that ensures equal gender probability when gender evidence is occluded in a scene and confident predictions when gender evidence is present. The resulting model is forced to look at a person rather than use contextual cues to make a gender-specific predictions. The losses that comprise our model, the Appearance Confusion Loss and the Confident Loss, are general, and can be added to any description model in order to mitigate impacts of unwanted bias in a description dataset. Our proposed model has lower error than prior work when describing images with people and mentioning their gender and more closely matches the ground truth ratio of sentences including women to sentences including men. We also show that unlike other approaches, our model is indeed more often looking at people when predicting their gender.