Structural transfer learning of non-Gaussian DAG
This addresses the challenge of pooling heterogeneous data for better DAG structure reconstruction in target studies, which is incremental as it builds on existing transfer learning methods but with a novel approach for non-Gaussian DAGs.
The paper tackles the problem of limited data for accurate directed acyclic graph (DAG) reconstruction by introducing a transfer learning framework that leverages heterogeneous data from multiple studies, showing substantial improvement in DAG reconstruction even when auxiliary DAGs are not overall similar.
Directed acyclic graph (DAG) has been widely employed to represent directional relationships among a set of collected nodes. Yet, the available data in one single study is often limited for accurate DAG reconstruction, whereas heterogeneous data may be collected from multiple relevant studies. It remains an open question how to pool the heterogeneous data together for better DAG structure reconstruction in the target study. In this paper, we first introduce a novel set of structural similarity measures for DAG and then present a transfer DAG learning framework by effectively leveraging information from auxiliary DAGs of different levels of similarities. Our theoretical analysis shows substantial improvement in terms of DAG reconstruction in the target study, even when no auxiliary DAG is overall similar to the target DAG, which is in sharp contrast to most existing transfer learning methods. The advantage of the proposed transfer DAG learning is also supported by extensive numerical experiments on both synthetic data and multi-site brain functional connectivity network data.