Synthetic Datasets for Low SNR Evaluation

该数据集是一组综合性的数据集，包括符号功能、图像和音频数据，旨在评估在低信噪比条件下的事后局部归因方法。此外，这些数据集被严格用于从零开始训练多种经典模型，并在多种条件下进行测试，包括不同的噪声水平和模型配置。该数据集的规模为10,000个样本，其任务是评估在低信噪比环境中的归因方法。

This is a comprehensive dataset collection including symbolic features, image and audio data, which is designed to evaluate post-hoc local attribution methods under low signal-to-noise ratio (SNR) conditions. Furthermore, this collection is strictly utilized to train various classic models from scratch and tested under diverse conditions, including different noise levels and model configurations. The dataset collection has a total of 10,000 samples, with its core task being the evaluation of attribution methods in low SNR environments.