GLASD combines adaptive stochastic descent with simulated-annealing-style exploration on a spherical parameterization of correlation matrices, but its global convergence guarantee is not actually proven.
An overview of the estimation of large covariance and precision matrices,
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
citation-role summary
background 1
citation-polarity summary
fields
stat.AP 1years
2025 1verdicts
REJECT 1roles
background 1polarities
background 1representative citing papers
citing papers explorer
-
GLASD: A Loss-Function-Agnostic Global Optimizer for Robust Correlation Estimation under Data Contamination and Heavy Tails
GLASD combines adaptive stochastic descent with simulated-annealing-style exploration on a spherical parameterization of correlation matrices, but its global convergence guarantee is not actually proven.