Identifying Network Hubs with the Partial Correlation Graphical LASSO
By: Małgorzata Bogdan , Adam Chojecki , Ivan Hejný and more
Potential Business Impact:
Finds hidden connections in complex data.
The Partial Correlation Graphical LASSO (PCGLASSO) offers a scale-invariant alternative to the standard GLASSO. This paper provides the first comprehensive treatment of the PCGLASSO estimator. We introduce a novel and highly efficient algorithm. Our central theoretical contribution is the first scale-invariant irrepresentability criterion for PCGLASSO, which guarantees consistent model selection. We prove this condition is significantly weaker than its GLASSO counterpart, providing the first theoretical justification for PCGLASSO's superior empirical performance, especially in recovering networks with hub structures. Furthermore, we deliver the first analysis of the estimator's non-convex solution landscape, establishing new conditions for global uniqueness and guaranteeing the consistency of all minimizers.
Similar Papers
Existence and optimisation of the partial correlation graphical lasso
Methodology
Finds hidden connections in data, even with little information.
Existence of the solution to the graphical lasso
Statistics Theory
Finds hidden connections in data, even when messy.
Modeling Large Nonstationary Spatial Data with the Full-Scale Basis Graphical Lasso
Methodology
Helps understand weather patterns from limited data.