Ergodic mirror descent
From MaRDI portal
Abstract: We generalize stochastic subgradient descent methods to situations in which we do not receive independent samples from the distribution over which we optimize, but instead receive samples that are coupled over time. We show that as long as the source of randomness is suitably ergodic---it converges quickly enough to a stationary distribution---the method enjoys strong convergence guarantees, both in expectation and with high probability. This result has implications for stochastic optimization in high-dimensional spaces, peer-to-peer distributed optimization schemes, decision problems with dependent data, and stochastic optimization problems over combinatorial spaces.
Recommendations
- On the convergence of mirror descent beyond stochastic convex programming
- Optimal distributed stochastic mirror descent for strongly convex optimization
- Primal-dual mirror descent method for constraint stochastic optimization problems
- Stochastic mirror descent method for distributed multi-agent optimization
- Algorithms of inertial mirror descent in convex problems of stochastic optimization
Cited in
(24)- On stochastic mirror descent with interacting particles: convergence properties and variance reduction
- Finite-sample analysis of nonlinear stochastic approximation with applications in reinforcement learning
- Hessian informed mirror descent
- Recursive aggregation of estimators by the mirror descent algorithm with averaging
- Event-triggered distributed online convex optimization with delayed bandit feedback
- Stochastic forward-backward splitting for monotone inclusions
- On the convergence of gradient-like flows with noisy gradient input
- Stochastic optimization for dynamic pricing
- Simple and optimal methods for stochastic variational inequalities. II: Markovian noise and policy evaluation in reinforcement learning
- Learning theory estimates with observations from general stationary stochastic processes
- Prediction of time series by statistical learning: general losses and fast rates
- On the convergence of mirror descent beyond stochastic convex programming
- Markov chain stochastic DCA and applications in deep learning with PDEs regularization
- Stochastic incremental mirror descent algorithms with Nesterov smoothing
- Parameter-free accelerated gradient descent for nonconvex minimization
- Bregman methods for large-scale optimization with applications in imaging
- Convergence rate bounds for the mirror descent method: IQCs, Popov criterion and Bregman divergence
- Parametric estimation of stochastic differential equations via online gradient descent
- Concentration of contractive stochastic approximation: additive and multiplicative noise
- Stochastic-constrained stochastic optimization with Markovian data
- Validation analysis of mirror descent stochastic approximation method
- Optimization with Markovian noise: towards optimal rates in strong growth case
- Inexact online proximal mirror descent for time-varying composite optimization
- Learning by mirror averaging
This page was built for publication: Ergodic mirror descent
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4915178)
