csu_synthetic_attribution

csu_synthetic_attribution数据集由科罗拉多州立大学创建，包含100万个样本，用于评估神经网络在地球科学问题中的归因方法。数据集通过使用可加性分离函数生成，确保每个样本的归因真实性已知，从而允许客观评估解释性人工智能（XAI）方法。该数据集适用于回归问题，特别是在气候预测等领域，旨在通过提供客观的归因基准来增强模型信任并辅助科学发现。

The csu_synthetic_attribution dataset, developed by Colorado State University, contains 1 million samples and is designed to evaluate attribution methods of neural networks for geoscientific applications. It is generated using additive separable functions, ensuring that the ground truth of attribution for each sample is known, thus enabling the objective evaluation of eXplainable Artificial Intelligence (XAI) methods. This dataset is suitable for regression tasks, particularly in domains such as climate prediction, and aims to enhance model trust and facilitate scientific discovery by providing an objective attribution benchmark.