Diffusion LMS Strategies for Distributed Estimation

By Federico S. Cattivelli, Ali H. Sayed. IEEE Transactions on Signal Processing, 2010

1,165 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Computational Mechanics, Signal Processing

View Paper: https://www.semanticscholar.org/paper/16bcb9f8b2742aa0b2222fec7fddf19157ff806d

How researchers see this paper

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

In Artificial Intelligence

In Computational Mechanics

In Signal Processing

Citations in the corpus

Adaptation, Learning, and Optimization over Networks (2014, background): “Additional details on the case C = IN appear in [62, 66, 209].”
Distributed Policy Evaluation Under Multiple Behavior Strategies (2013, methodology): “We extend the energy conservation arguments of [16]– [19] to perform a mean-square-error (MSE) analysis of the diffusion GTD algorithm (33a)–(33d) and provide convergence guarantees under sufficiently small step-sizes.”
Performance of a Distributed Stochastic Approximation Algorithm (2012, positive): “The convergence results rely on the following equations: under Assumption 1a) , it holds We then ﬁrst address the almost-sure convergence of the sequence ) n converges almost-surely to zero and, (ii) by applying results on the convergence of stochastic approximation algorithms with state-dependent…”
Diffusion Adaptation Strategies for Distributed Optimization and Learning Over Networks (2011, methodology): “If we use the instantaneous approximations as is common in the context of stochastic approximation and adaptive filtering [44], [46], [52], then the actual gradient ∇wJl(w) can be approximated by ∇̂wJl(w) = ∇wQl(w, {ul,i,dl(i)}) = −2ul,i[dl(i)− ul,iw] (56) Substituting into (24)–(25), and assuming C”
Diffusion Adaptation over Networks (2012, methodology): “While we can continue with the analysis by taking this factor into account, as was done in [4, 5, 18, 57], it is sufficient for the exposition in this article to focus on the case of sufficiently small step-sizes where terms involving higher powers of the step-sizes can be ignored.”
Adaptive Networks (2014, methodology): “The ATC diffusion structure (49b), with adaptation preceding combination, appeared in the work [114] on adaptive distributed least squares schemes and also in the works [98] and [115]–[117] on distributed MSE and state–space estimation methods.”
Diffusion Strategies for Distributed Kalman Filtering and Smoothing (2010, background): “Our work is motivated by our earlier realization that while consensus solutions are useful for computing averages, diffusion solutions are particularly suited for problems involving the recursive minimization of cost functions [16], [21]–[23], [27]–[29].”
Robust Distributed Estimation by Networked Agents (2017, methodology): “Traditionally in the energy conservation framework [8], [11], [20], one evaluates the weighted variance relation associated with (64) by equating the squared weighted Euclidean norms of both sides with respect to a symmetric nonnegative-definite weighting matrix Σ that we are free to choose, taking ”
Multitask Diffusion Adaptation Over Networks (2013, neutral): “Boldface capital letters denote matrices.”
Diffusion strategies for adaptation and learning over networks: an examination of distributed strategies and network behavior (2013, methodology): “Different choices for R enable us to extract different types of information about the performance of the agents and the network [15], [16].”
Diffusion Least-Mean Squares With Adaptive Combiners: Formulation and Performance Analysis (2010, methodology): “In [14], performance analysis of more general CTA and ATC diffusion LMS algorithms equipped with static combiners has been done.”
Network Topology and Communication-Computation Tradeoffs in Decentralized Optimization (2017, background): “…to assume that the observations { ζ i } ni =1 are conditionally independent given the model parameters x . of In this case, the maximum likelihood estimate x can obtained by solving which can be addressed using multi-agent consensus optimization methods [4]–[6] with f i ( x ) = log p ( ζ i | x ) .”