20 synthetic datasets

该数据集汇总了从已知网络中抽取的合成数据集，并使用了BNgenerator软件包进行生成。其中五个数据集来源于文献资料，另外十五个则由作者自行生成。该数据集规模宏大，包含多达10000个变量，每个训练数据集包含5000个实例。其任务是对结构学习及算法进行比较研究。

This dataset collection compiles synthetic datasets extracted from known Bayesian networks and generated using the BNgenerator software package. Specifically, five of these datasets are sourced from existing literature, while the other fifteen were independently generated by the authors. This collection has a considerable scale, with each dataset containing up to 10,000 variables, and each training dataset comprising 5,000 instances. Its primary purpose is to support comparative research on structure learning and associated algorithms.