CoLiDE: Concomitant Linear DAG Estimation

arXiv:2310.02895v2 Announce Type: replace
Abstract: We deal with the combinatorial problem of learning directed acyclic graph (DAG) structure from observational data adhering to a linear structural equation model (SEM). Leveraging advances in differentiable, nonconvex characterizations of acyclicity, recent efforts have advocated a continuous constrained optimization paradigm to efficiently explore the space of DAGs. Most existing methods employ lasso-type score functions to guide this search, which (i) require expensive penalty parameter retuning when the $\textit{unknown}$ SEM noise variances change across problem …

abstract advances arxiv continuous cs.lg dag data deal differentiable equation explore graph linear optimization paradigm sem space type