Multitask Diffusion Adaptation OverAsynchronous Networks

Abstract

Abstract-The multitask diffusion LMS is an efficient strategy to simultaneously infer, in a collaborative manner, multiple parameter vectors. Existing works on multitask problems assume that all agents respond to data synchronously. In several applications, agents may not be able to act synchronously because networks can be subject to several sources of uncertainties such as changing topology, random link failures, or agents turning on and off for energy conservation. In this work, we describe a model for the solution of multitask problems over asynchronous networks and carry out a detailed mean and mean-square error analysis. Results show that sufficiently small step-sizes can still ensure both stability and performance. Simulations and illustrative examples are provided to verify the theoretical findings.

Transient analysis of multitask learning over adaptive networks with wireless links

“…[2][3][4][5][6][7][8][9][10][11] Adaptive networks are split into multi-task and single-task networks. [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] In a single-task network, there is one target vector to estimate; however, in a multi-task network, multiple target vectors should be estimated. [11][12][13][14][15][16][17] The links in the topology of adaptive networks can be connected through wireless or wired links.…”

“…[1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17] In a single-task network, there is one target vector to estimate; however, in a multi-task network, multiple target vectors should be estimated. [11][12][13][14][15][16][17] The links in the topology of adaptive networks can be connected through wireless or wired links. In References 18-23 the learning performance of distributed algorithms for single-task networks with noisy wired link, and in References 24-28, the learning performance of distributed algorithms for single-task networks with wireless fading links were studied.…”

“…There are several policies for the type of cooperation between the agents where the diffusion policy is the best policy given the scalability and robustness 2‐11 . Adaptive networks are split into multi‐task and single‐task networks 1‐17 . In a single‐task network, there is one target vector to estimate; however, in a multi‐task network, multiple target vectors should be estimated 11‐17 …”

“…Adaptive networks are split into multi‐task and single‐task networks 1‐17 . In a single‐task network, there is one target vector to estimate; however, in a multi‐task network, multiple target vectors should be estimated 11‐17 …”

Adaptive diffusion strategies with Markov jump over networks

“…There exist several useful techniques for the solution of optimization problems in a distributed manner. Among them, the most representative are incremental methods (Lopes & Sayed, 2007;Nedic & Bertsekas, 2001;Rabbat & Nowak, 2005), consensus methods (Chouvardas et al, 2011;Kar & Moura, 2011;Schizas et al, 2008) and diffusion strategies (Abdolee et al, 2016;Chen & Sayed, 2012a;Eltokhey et al, 2018;Matta et al, 2016;Nassif et al, 2016;Sayed, 2014). Other research references are available (Chen et al, 2018;Chen & Shao, 2017).…”