Adaptive online distributed optimization in dynamic environments

Nazari, Parvin; Khorram, Esmaeil; Tarzanagh, Davoud Ataee

doi:10.1080/10556788.2019.1637433

Cited by 13 publications

(9 citation statements)

References 26 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, the momentum acceleration technique was exploited to improve the performance under time-varying unbalanced communication graphs in [128], where an improved static regret bound O( √ 1 + log T + √ T ) is established for convex cost functions. Also, adaptive gradient methods have been integrated into DOL in [129]- [131].…”

Section: Metricmentioning

confidence: 99%

“…The most popular adaptive gradient method is Adam by estimating first-and second-order moments of gradients [139]. As a result, one natural idea is to apply adaptive methods to DOL for improving the performance, which is exactly done in [129]- [131]. However, the research along this line is not far fully explored, leaving an enormous possibility for future directions.…”

Section: Future Directionsmentioning

confidence: 99%

See 1 more Smart Citation

A Survey on Distributed Online Optimization and Game

Li¹,

Xie²,

Li³

2022

Preprint

View full text Add to dashboard Cite

Decentralized online learning (DOL) has been increasingly researched in the last decade, mostly motivated by its wide applications in sensor networks, commercial buildings, robotics (e.g., decentralized target tracking and formation control), smart grids, deep learning, and so forth. In this problem, there are a network of agents who may be cooperative (i.e., decentralized online optimization) or noncooperative (i.e., online game) through local information exchanges, and the local cost function of each agent is often time-varying in dynamic and even adversarial environments. At each time, a decision must be made by each agent based on historical information at hand without knowing future information on cost functions. Although this problem has been extensively studied in the last decade, a comprehensive survey is lacking. Therefore, this paper provides a thorough overview of DOL from the perspective of problem settings, communication, computation, and performances. In addition, some potential future directions are also discussed in details.

show abstract

Section: Metricmentioning

confidence: 99%

Section: Future Directionsmentioning

confidence: 99%

A Survey on Distributed Online Optimization and Game

Li¹,

Xie²,

Li³

2022

Preprint

View full text Add to dashboard Cite

show abstract

“…Comparing with the SGD method, ADAGRAD dynamically interpolates knowledge of history gradients to adaptively change the learning rate, thereby achieving significantly better performance when the gradients are sparse, or in general small. Another state-of-the-art algorithm for training deep learning models is known as ADAM [18,29], which is a variant of the general class of ADAGRAD-type algorithms. When 0 < β 2 < 1, 0 ≤…”

Section: Adagrad and Adam For Minimization Problemsmentioning

confidence: 99%

Dynamic Regret Analysis for Online Meta-Learning

Parvin¹,

Khorram²

2021

Preprint

Self Cite

View full text Add to dashboard Cite

The online meta-learning framework has arisen as a powerful tool for the continual lifelong learning setting. The goal for an agent is to quickly learn new tasks by drawing on prior experience, while it faces with tasks one after another. This formulation involves two levels: outer level which learns meta-learners and inner level which learns task-specific models, with only a small amount of data from the current task. While existing methods provide static regret analysis for the online meta-learning framework, we establish performance in terms of dynamic regret which handles changing environments from a global prospective. We also build off of a generalized version of the adaptive gradient methods that covers both ADAM and ADAGRAD to learn meta-learners in the outer level. We carry out our analyses in a stochastic setting, and in expectation prove a logarithmic local dynamic regret which depends explicitly on the total number of iterations T and parameters of the learner. Apart from, we also indicate high probability bounds on the convergence rates of proposed algorithm with appropriate selection of parameters, which have not been argued before.

show abstract

“…To overcome this issue, several improved variants of SGD that automatically update the search directions and learning rates using a metric constructed from the history of iterates have been proposed. The pioneering work in this line of research is [23], while recent developments include [12,47,43,24,11,36,33,32,31].…”

Section: Introductionmentioning

confidence: 99%

Adaptive First-and Zeroth-order Methods for Weakly Convex Stochastic Optimization Problems

Parvin¹,

Tarzanagh²,

Michailidis³

2020

Preprint

Self Cite

View full text Add to dashboard Cite

In this paper, we design and analyze a new family of adaptive subgradient methods for solving an important class of weakly convex (possibly nonsmooth) stochastic optimization problems. Adaptive methods which use exponential moving averages of past gradients to update search directions and learning rates have recently attracted a lot of attention for solving optimization problems that arise in machine learning. Nevertheless, their convergence analysis almost exclusively requires smoothness and/or convexity of the objective function. In contrast, we establish non-asymptotic rates of convergence of first and zeroth-order adaptive methods and their proximal variants for a reasonably broad class of nonsmooth & nonconvex optimization problems. Experimental results indicate how the proposed algorithms empirically outperform stochastic gradient descent and its zeroth-order variant for solving such optimization problems.

show abstract

Adaptive online distributed optimization in dynamic environments

Cited by 13 publications

References 26 publications

A Survey on Distributed Online Optimization and Game

A Survey on Distributed Online Optimization and Game

Dynamic Regret Analysis for Online Meta-Learning

Adaptive First-and Zeroth-order Methods for Weakly Convex Stochastic Optimization Problems

Contact Info

Product

Resources

About