Diffusing DeBias: Synthetic Bias Amplification for Model Debiasing
It addresses bias in deep learning models for classification tasks, offering a novel plug-in approach that improves generalization, though it is incremental as it builds on existing debiasing techniques.
The paper tackles bias in deep learning models caused by spurious correlations in training data by introducing Diffusing DeBias (DDB), a method that uses diffusion models to generate synthetic bias-aligned images for debiasing, achieving state-of-the-art results on multiple benchmark datasets.
Deep learning model effectiveness in classification tasks is often challenged by the quality and quantity of training data whenever they are affected by strong spurious correlations between specific attributes and target labels. This results in a form of bias affecting training data, which typically leads to unrecoverable weak generalization in prediction. This paper aims at facing this problem by leveraging bias amplification with generated synthetic data: we introduce Diffusing DeBias (DDB), a novel approach acting as a plug-in for common methods of unsupervised model debiasing exploiting the inherent bias-learning tendency of diffusion models in data generation. Specifically, our approach adopts conditional diffusion models to generate synthetic bias-aligned images, which replace the original training set for learning an effective bias amplifier model that we subsequently incorporate into an end-to-end and a two-step unsupervised debiasing approach. By tackling the fundamental issue of bias-conflicting training samples memorization in learning auxiliary models, typical of this type of techniques, our proposed method beats current state-of-the-art in multiple benchmark datasets, demonstrating its potential as a versatile and effective tool for tackling bias in deep learning models. Code is available at https://github.com/Malga-Vision/DiffusingDeBias