Variance Reduction via Primal-Dual Accelerated Dual Averaging for Nonsmooth Convex Finite-Sums.
Chaobing SongStephen J. WrightJelena DiakonikolasPublished in: CoRR (2021)
Keyphrases
- primal dual
- variance reduction
- convex functions
- convex optimization
- variational inequalities
- convex programming
- linear programming
- monte carlo
- linear program
- sample size
- convex optimization problems
- interior point methods
- linear programming problems
- infeasible interior point
- approximation algorithms
- semidefinite programming
- saddle point
- algorithm for linear programming
- convergence rate
- duality gap
- dual formulation
- importance sampling
- naive bayes classifier
- convex sets
- machine learning
- augmented lagrangian
- finite number
- random variables
- newton method
- confidence intervals
- total variation
- image restoration
- naive bayes
- image processing