Common-Sense Bias Modeling for Classification Tasks

TL;DR

This paper introduces a framework that uses textual descriptions to identify and mitigate common sense-related biases in image datasets, improving model robustness by addressing complex feature correlations.

Contribution

It presents a novel method for extracting comprehensive biases from image datasets using semantic clustering of noun phrases and human-in-the-loop analysis.

Findings

Uncovers new biases in multiple image benchmarks.

Data re-weighting effectively reduces identified biases.

Outperforms existing bias mitigation techniques.

Abstract

Machine learning model bias can arise from dataset composition: correlated sensitive features can distort the downstream classification model's decision boundary and lead to performance differences along these features. Existing de-biasing works tackle the most prominent bias features, such as colors of digits or background of animals. However, real-world datasets often include a large number of feature correlations that intrinsically manifest in the data as common sense information. Such spurious visual cues can further reduce model robustness. Thus, domain practitioners desire a comprehensive understanding of correlations and the flexibility to address relevant biases. To this end, we propose a novel framework to extract comprehensive biases in image datasets based on textual descriptions, a common sense-rich modality. Specifically, features are constructed by clustering noun phrase embeddings with similar semantics. The presence of each feature across the dataset is inferred, and their co-occurrence statistics are measured, with spurious correlations optionally examined by a human-in-the-loop module. Downstream experiments show that our method uncovers novel model biases in multiple image benchmark datasets. Furthermore, the discovered bias can be mitigated by simple data re-weighting to de-correlate the features, outperforming state-of-the-art unsupervised bias mitigation methods.