An efficient algorithm for large-scale causal discovery

作者：Hong, Yinghan^*; Liu, Zhusong; Mai, Guizhen

来源：Soft Computing, 2017, 21(24): 7381-7391.

DOI：10.1007/s00500-016-2281-0

• Causation discovery
• Causal network
• Additive noise model

Causal discovery is a fundamental problem in scientific research. Although many researchers are committed to finding causal relationships from observational data, large-scale causal discovery remains a tremendous challenge. In this paper, a new approach for large-scale causal discovery is proposed, based on a split-and-merge strategy. The method first splits a given dataset into small subdatasets using a graph-partitioning method and then develops a effective algorithm to infer the causality of each subdataset. The entire causal structure with respect to the given dataset is achieved by combining all the causalities of each subdataset. The experimental results show that the proposed approach is effective and scalable for large-scale causal discovery problems.

• 出版日期2017-12
• 单位广东工业大学; 韩山师范学院