Debiasing Classifiers by Amplifying Bias with Latent Diffusion and Large Language Models
This addresses the issue of biased classifiers for researchers and practitioners in computer vision, though it is incremental as it builds on existing generative models.
The paper tackles the problem of neural networks learning spurious correlations in image classification by introducing DiffuBias, a pipeline that generates bias-conflict samples using pretrained diffusion and image captioning models, achieving state-of-the-art performance on benchmark datasets.
Neural networks struggle with image classification when biases are learned and misleads correlations, affecting their generalization and performance. Previous methods require attribute labels (e.g. background, color) or utilizes Generative Adversarial Networks (GANs) to mitigate biases. We introduce DiffuBias, a novel pipeline for text-to-image generation that enhances classifier robustness by generating bias-conflict samples, without requiring training during the generation phase. Utilizing pretrained diffusion and image captioning models, DiffuBias generates images that challenge the biases of classifiers, using the top-$K$ losses from a biased classifier ($f_B$) to create more representative data samples. This method not only debiases effectively but also boosts classifier generalization capabilities. To the best of our knowledge, DiffuBias is the first approach leveraging a stable diffusion model to generate bias-conflict samples in debiasing tasks. Our comprehensive experimental evaluations demonstrate that DiffuBias achieves state-of-the-art performance on benchmark datasets. We also conduct a comparative analysis of various generative models in terms of carbon emissions and energy consumption to highlight the significance of computational efficiency.