面向多标签图像分类的基于原型的因果干预
Prototype-based Causal Intervention for Multi-Label Image Classification
用可学习的上下文原型在记忆库中动态建模混淆因子，仅凭图像级监督近似后门调整，为视觉分类中去除共现偏差提供了一个可复用的因果干预模块。
原文依据 approximates the backdoor adjustment using only image-level supervision
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论现代多标签图像分类模型严重依赖虚假相关，无法学到潜在的因果机制。许多受因果启发的方���不切实际，需要现实数据集中很少提供的框级监督。另一些方法依赖静态混淆因子字典，这种字典本质上缺乏灵活性，既无法捕捉复杂偏差，也无法适应训练过程中的特征空间变化。为解决这一问题，我们提出基于原型的因果干预（ProCI），这是一个仅使用图像级监督来近似后门调整的新框架。它将混淆因子建模为可学习的上下文原型，这些原型被设计用来表示各类别的共现偏差。这些原型在稳定的记忆库中动态学习，并被用来构造样本特定的偏差向量以进行自适应特征调整，从而有效抵消虚假相关。在 MS-COCO、Pascal VOC、COCO-Stuff 以及具有挑战性的 Sewer-ML 数据集上的实验验证了我们的方法。ProCI 在标准基准上取得了有竞争力的性能，同时在高度混杂的 Sewer-ML 上创造了新的最优结果。它在 F2-CIW 指标上以显著的 +5.44 点超越了此前的最佳模型。这些结果表明，我们的方法仅使用图像级监督就能有效缓解复杂的现实世界偏差。
英文原文摘要 1403 字符
Modern multi-label image classification models suffer from a critical reliance on spurious correlations, failing to learn the underlying causal mechanisms. Many causality-inspired methods are impractical, demanding box-level supervision that is rarely available in real-world datasets. Others rely on static confounder dictionaries, which are inherently inflexible and fail to capture complex biases or adapt to feature space changes during training. To address this, we present prototype-based causal intervention (ProCI), a novel framework that approximates the backdoor adjustment using only image-level supervision. It models confounders as learnable contextual prototypes engineered to represent class-wise co-occurring bias. These prototypes are learned dynamically within a stable memory and leveraged to construct sample-specific bias vectors for an adaptive feature adjustment, effectively counteracting spurious correlations. Experiments on MS-COCO, Pascal VOC, COCO-Stuff, and the challenging Sewer-ML dataset validate our approach. ProCI achieves competitive performance on standard benchmarks while setting a new state-of-the-art on the highly-confounded Sewer-ML. It outperforms the previous best model by a remarkable +5.44 points on the F2-CIW metric. These results demonstrate the effectiveness of our approach in mitigating complex real-world biases using only image-level supervision.