Diffusion Least-Mean Squares Over Adaptive Networks: Formulation and Performance Analysis

By Cássio Guimarães Lopes, Ali H. Sayed. IEEE Transactions on Signal Processing, 2008

1,108 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computer Networks and Communications

View Paper: https://www.semanticscholar.org/paper/c7b572ca4b537d709287e1f4943ed7912a770c0d

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

In Computer Networks and Communications

Citations in the corpus

Diffusion Least-Mean Squares With Adaptive Combiners: Formulation and Performance Analysis (2010, methodology): “For example, using the LMS algorithm as the core adaptive filter leads to the CTA diffusion LMS algorithm [10]:”
Decentralized double stochastic averaging gradient (2015, background): “Analogous formulations are also of interest in decentralized control [3], [4] and sensor networks [5]–[7].”
Adaptation, Learning, and Optimization over Networks (2014, background): “22) was first introduced for mean-square-error estimation problems in [160, 161, 164, 212].”
DQM: Decentralized Quadratically Approximated Alternating Direction Method of Multipliers (2015, background): “The nodes’ goal is to ﬁnd the optimal argument of the global cost function Problems of this form arise in, e.g., decentralized control [2]– [4], wireless communication [5], [6], sensor networks [7]–[9], and large scale machine learning [10]–[12].”
Distributed Policy Evaluation Under Multiple Behavior Strategies (2013, methodology): “There are sev eral distributed strategies that can be used for this purpose, such as consensus [10]–[14] and diffusion trategies [15]–[18].”
Adaptive Networks (2014, methodology): “The CTA diffusion strategy (49a) was first introduced for MSE estimation problems in [97] and [110]–[112].”
Multitask Diffusion Adaptation Over Networks (2013, background): “Although centralized strategies with a fusion center can beneﬁt more fully from information collected throughout the network but stored at a single point, in most cases, distributed strategies are more attractive to solve inference problems in a collaborative and autonomous manner.”
Diffusion strategies for adaptation and learning over networks: an examination of distributed strategies and network behavior (2013, methodology): “C I = The CTA diffusion strategy (14) was proposed in [32]–[34] and [28], and the ATC diffusion structure (16), with adaptation preceding combination, was proposed in the work [35] on distributed least-squares schemes and subsequently in the works [36]–[38] and [16] on distributed MSE and state-spac”
Diffusion Strategies for Distributed Kalman Filtering and Smoothing (2010, background): “Convex combinations as in (19) have been considered before in the context of adaptive filtering [16], [22], [30]–[32].”
Gossip Algorithms for Distributed Signal Processing (2010, background): “An equivalent formulation of the estimation problem in the setting considered above, comes from the distributed least mean square (LMS) adaptive filtering framework [92]–[94].”
A Framework for Parallel and Distributed Training of Neural Networks (2016, methodology): “We have selected the weight coefficients in (2) using the MetropolisHastings strategy (Lopes and Sayed, 2008): Cij = 1 max{δi,δj}+1 i 6= j, j ∈ Ni 1− ∑ j∈Ni 1 max{δi,δj}+1 i = j 0 otherwise (37) where δi is the degree of node i.”
Distributed LMS for Consensus-Based In-Network Adaptive Processing (2009, background): “However, since is typically much smaller than the increase in computational complexity is relatively low.”
Distributed Parameter Estimation in Sensor Networks: Nonlinear Observation Models and Imperfect Communication (2008, methodology): “The use of non-decaying combining weights in [22], [24], [25] le ad to a residual error, however, under appropriate assumptions, these algorithms can be adapted for tracking c ertain time-varying parameters.”
Decentralized RLS With Data-Adaptive Censoring for Regressions Over Large-Scale Networks (2016, background): “We are interested in the global mean-square deviation (GMSD) metric [12], [22], defined as”
Diffusion LMS Strategies for Distributed Estimation (2010, methodology): “in [7], while a newer, more powerful adaptive strategy was proposed in [24].”