Stochastic variance reduction
ID: stochastic-variance-reduction
Stochastic Variance Reduction is a collection of techniques used in optimization and statistical estimation to reduce the variance of estimators or gradients when dealing with stochastic or noisy data. The goal is to achieve better convergence rates and more stable estimates in stochastic optimization problems, particularly in the context of algorithms such as stochastic gradient descent (SGD).
New to topics? Read the docs here!