Anisotropic Gaussian smoothing with step-dependent covariance matrices is inserted into GD, SGD, and Adam, and convergence bounds are derived that generalize the isotropic case.
Completely derandomized self- adaptation in evolution strategies
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
math.OC 1years
2024 1verdicts
REJECT 1representative citing papers
citing papers explorer
-
Anisotropic Gaussian Smoothing for Gradient-based Optimization
Anisotropic Gaussian smoothing with step-dependent covariance matrices is inserted into GD, SGD, and Adam, and convergence bounds are derived that generalize the isotropic case.