pith. sign in

arxiv: 1607.01027 · v5 · pith:Y5ZCMVY6new · submitted 2016-07-04 · 🧮 math.OC · cs.LG· cs.NA· math.NA· stat.ML

Accelerate Stochastic Subgradient Method by Leveraging Local Growth Condition

classification 🧮 math.OC cs.LGcs.NAmath.NAstat.ML
keywords stochasticgrowthlocalmathbfoptimalratesolutionsubgradient
0
0 comments X
read the original abstract

In this paper, a new theory is developed for first-order stochastic convex optimization, showing that the global convergence rate is sufficiently quantified by a local growth rate of the objective function in a neighborhood of the optimal solutions. In particular, if the objective function $F(\mathbf w)$ in the $\epsilon$-sublevel set grows as fast as $\|\mathbf w - \mathbf w_*\|_2^{1/\theta}$, where $\mathbf w_*$ represents the closest optimal solution to $\mathbf w$ and $\theta\in(0,1]$ quantifies the local growth rate, the iteration complexity of first-order stochastic optimization for achieving an $\epsilon$-optimal solution can be $\widetilde O(1/\epsilon^{2(1-\theta)})$, which is optimal at most up to a logarithmic factor. To achieve the faster global convergence, we develop two different accelerated stochastic subgradient methods by iteratively solving the original problem approximately in a local region around a historical solution with the size of the local region gradually decreasing as the solution approaches the optimal set. Besides the theoretical improvements, this work also includes new contributions towards making the proposed algorithms practical: (i) we present practical variants of accelerated stochastic subgradient methods that can run without the knowledge of multiplicative growth constant and even the growth rate $\theta$; (ii) we consider a broad family of problems in machine learning to demonstrate that the proposed algorithms enjoy faster convergence than traditional stochastic subgradient method. We also characterize the complexity of the proposed algorithms for ensuring the gradient is small without the smoothness assumption.

This paper has not been read by Pith yet.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Stochastic algorithms with geometric step decay converge linearly on sharp functions

    math.OC 2019-07 unverdicted novelty 7.0

    Geometric step decay yields local linear convergence for stochastic algorithms on sharp nonconvex problems and gives matching or new guarantees for phase retrieval and blind deconvolution under Gaussian and heavy-tail...