DiffInject：通过基于扩散的风格注入重新审视合成数据生成去偏方法
DiffInject: Revisiting Debias via Synthetic Data Generation using Diffusion-based Style Injection
AI总结：
本文提出 DiffInject，利用预训练扩散模型操纵潜在空间生成偏差冲突样本，在完全无监督条件下有效缓解数据集偏差。
摘要：
数据集偏差是机器学习中的一个重大挑战，其中图像的纹理或颜色等特定属性会被无意中学习，从而导致性能受损。为解决这一问题，以往的工作主要集中在通过开发新的去偏算法或生成合成数据来缓解普遍存在的数据集偏差。然而，迄今为止的生成式方法在很大程度上依赖于使用数据集中特定于偏差的样本，而这些样本通常过于稀缺。在这项工作中，我们提出了 DiffInject，一种简单而有效的方法，利用预训练的扩散模型来增广合成偏差冲突样本。该方法通过操纵潜在空间，显著推动了扩散模型在去偏方面的应用。我们的框架不需要任何关于偏差类型或标签的显式知识，使其成为一种完全无监督的去偏设置。我们的方法在有效减少数据集偏差方面展示了显著的结果。
英文摘要：
Dataset bias is a significant challenge in machine learning, where specific attributes, such as texture or color of the images are unintentionally learned resulting in detrimental performance. To address this, previous efforts have focused on debiasing models either by developing novel debiasing algorithms or by generating synthetic data to mitigate the prevalent dataset biases. However, generative approaches to date have largely relied on using bias-specific samples from the dataset, which are typically too scarce. In this work, we propose, DiffInject, a straightforward yet powerful method to augment synthetic bias-conflict samples using a pretrained diffusion model. This approach significantly advances the use of diffusion models for debiasing purposes by manipulating the latent space. Our framework does not require any explicit knowledge of the bias types or labelling, making it a fully unsupervised setting for debiasing. Our methodology demonstrates substantial result in effectively reducing dataset bias.
机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。
