COARSE: Score-Based Learning of Cluster DAGs
What it is
COARSE is a score-based method for learning high-level causal cluster DAGs from interventional data. It uses a two-phase approach where the first phase identifies clusters and the second phase recovers edges using a local search under a cluster-level BIC score.
The problem it solves
Existing constraint-based methods for learning cluster DAGs are computationally expensive and scale poorly to dense graphs with many variables.
What you could use it for
- Learning interpretable high-level causal models from complex low-level variable sets
- Recovering causal structures in systems with hundreds of variables using interventional data
Read the source ↗Back to the library →
Summary written by an AI model from the source; check the source before you rely on it.