Unsupervised Concept Discovery Mitigates Spurious Correlations
This addresses the issue of spurious correlations in machine learning models, which can cause unintended biases and unreliable predictions, particularly in applications where group annotations are unavailable, offering a novel unsupervised approach.
The paper tackles the problem of models being prone to spurious correlations in training data, which leads to brittle predictions and biases, by introducing CoBalT, a concept balancing technique that mitigates these correlations without requiring human labeling of subgroups, achieving superior or competitive performance on benchmark datasets for sub-population shifts.
Models prone to spurious correlations in training data often produce brittle predictions and introduce unintended biases. Addressing this challenge typically involves methods relying on prior knowledge and group annotation to remove spurious correlations, which may not be readily available in many applications. In this paper, we establish a novel connection between unsupervised object-centric learning and mitigation of spurious correlations. Instead of directly inferring subgroups with varying correlations with labels, our approach focuses on discovering concepts: discrete ideas that are shared across input samples. Leveraging existing object-centric representation learning, we introduce CoBalT: a concept balancing technique that effectively mitigates spurious correlations without requiring human labeling of subgroups. Evaluation across the benchmark datasets for sub-population shifts demonstrate superior or competitive performance compared state-of-the-art baselines, without the need for group annotation. Code is available at https://github.com/rarefin/CoBalT.