A Generic online acceleration scheme for Optimization algorithms via Relaxation and Inertia
arXiv:1603.05398
Abstract
We propose generic acceleration schemes for a wide class of optimization and iterative schemes based on relaxation and inertia. In particular, we introduce methods that automatically tunes the acceleration coefficients online, and establish their convergence. This is made possible by considering the class of fixed-points iterations over averaged operators which encompass gradient methods, ADMM, primal dual algorithms, an so on.
References in corpus (11)
- On the Linear Convergence of the ADMM in Decentralized Consensus Optimization
- An inertial forward-backward algorithm for monotone inclusions
- The rate of convergence of Nesterov's accelerated forward-backward method is actually faster than
- From Averaging to Acceleration, There is Only a Step-size
- Analysis of Sum-Weight-like algorithms for averaging in Wireless Sensor Networks
- Asynchronous Distributed Optimization using a Randomized Alternating Direction Method of Multipliers
- Optimal rates of convergence of matrices with applications
- Local Linear Convergence of ISTA and FISTA on the LASSO Problem
- Line Search for Averaged Operator Iteration
- Local and Global Convergence of an Inertial Version of Forward-Backward Splitting
- Smooth Strongly Convex Interpolation and Exact Worst-case Performance of First-order Methods