Towards Measuring Bias in Image Classification

TL;DR

This paper introduces a systematic method using attribution maps and metrics to detect and measure data bias in CNN image classification models, enhancing understanding of biases in complex neural networks.

Contribution

It presents a novel approach combining attribution maps and metrics to effectively uncover and quantify data bias in CNNs, validated on a synthetic biased dataset.

Findings

Some attribution map techniques better highlight data bias.

Metrics can effectively measure the bias in attribution maps.

The approach helps identify biases that are otherwise hard to detect.

Abstract

Convolutional Neural Networks (CNN) have become de fact state-of-the-art for the main computer vision tasks. However, due to the complex underlying structure their decisions are hard to understand which limits their use in some context of the industrial world. A common and hard to detect challenge in machine learning (ML) tasks is data bias. In this work, we present a systematic approach to uncover data bias by means of attribution maps. For this purpose, first an artificial dataset with a known bias is created and used to train intentionally biased CNNs. The networks' decisions are then inspected using attribution maps. Finally, meaningful metrics are used to measure the attribution maps' representativeness with respect to the known bias. The proposed study shows that some attribution map techniques highlight the presence of bias in the data better than others and metrics can support the identification of bias.