Provably Accelerated Decentralized Gradient Method Over Unbalanced Directed Graphs
From MaRDI portal
Abstract: In this work, we consider the decentralized optimization problem in which a network of agents, each possessing a smooth and convex objective function, wish to collaboratively minimize the average of all the objective functions through peer-to-peer communication in a directed graph. To solve the problem, we propose two accelerated Push-DIGing methods termed APD and APD-SC for minimizing non-strongly convex objective functions and strongly convex ones, respectively. We show that APD and APD-SC respectively converge at the rates and up to constant factors depending only on the mixing matrix. To the best of our knowledge, APD and APD-SC are the first decentralized methods to achieve provable acceleration over unbalanced directed graphs. Numerical experiments demonstrate the effectiveness of both methods.
Recommendations
- A decentralized Nesterov gradient method for stochastic optimization over unbalanced directed networks
- Balancing communication and computation in gradient tracking algorithms for decentralized optimization
- A distributed accelerated optimization algorithm over time‐varying directed graphs with uncoordinated step‐sizes
- Towards accelerated rates for distributed optimization over time-varying networks
- Distributed consensus-based multi-agent convex optimization via gradient tracking technique
Cited in
(3)
This page was built for publication: Provably Accelerated Decentralized Gradient Method Over Unbalanced Directed Graphs
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6373744)
