Optimal combination rules for adaptation and learning over networks

Tu, Sheng-Yuan; Sayed, Ali H.

doi:10.1109/camsap.2011.6136014

Cited by 51 publications

(78 citation statements)

References 8 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We pursue an approximate solution that relies on optimizing an upper bound and performs well in practice. It is shown in [14], [15] that the network MSD (52) is upper bounded by…”

Section: Optimizing the Combination Matricesmentioning

confidence: 99%

See 1 more Smart Citation

Combination weights for diffusion strategies with imperfect information exchange

Zhao

Sayed

2012

2012 IEEE International Conference on Communications (ICC)

Self Cite

View full text Add to dashboard Cite

Adaptive networks rely on in-network and collaborative processing among distributed agents to deliver enhanced performance in estimation and inference tasks. Information is exchanged among the nodes, usually over noisy links. This paper first investigates the mean-square performance of adaptive diffusion algorithms in the presence of various sources of imperfect information exchanges and quantization errors. Among other results, the analysis reveals that link noise over the regression data modifies the dynamics of the network evolution, and leads to biased estimates in steady-state. The analysis also reveals how the network mean-square performance is dependent on the combination weight matrices. We use these observations to show how the combination weights can be optimized and adapted. Simulation results illustrate the theoretical findings and match well with theory.

show abstract

“…We pursue an approximate solution that relies on optimizing an upper bound and performs well in practice. It is shown in [14], [15] that the network MSD (52) is upper bounded by…”

Section: Optimizing the Combination Matricesmentioning

confidence: 99%

“…(57) We refer to this combination rule as the relative-variance combination rule; it is an extension of the rule devised in [15] to the case of noisy information exchanges. Minimizing the upper bound of the network EMSE for the ATC algorithm over left-stochastic matrices A leads to the same solution (57).…”

Section: Optimizing the Combination Matricesmentioning

confidence: 99%

Combination weights for diffusion strategies with imperfect information exchange

Zhao

Sayed

2012

2012 IEEE International Conference on Communications (ICC)

Self Cite

View full text Add to dashboard Cite

show abstract

“…Optimal combination rule was suggested to fuse the information from the neighbor (10) . The rule considers the variation in noise profile across the nodes, while other structurally defined combination methods such as uniform rule, metropolis rule ignore the variance so that can lead to the performance degradation.…”

Section: Optimal Combination Rulementioning

confidence: 99%

“…The optimal combination rule (10) , uniform rule, and combination method based MSD (4) were compared with the proposed algorithm in the same simulation environment. The uniform rule is as follows…”

Section: Simulationmentioning

confidence: 99%

“…They consider the number of the linked nodes, but not other properties. As the convergence rate and steady-state error depend on the various noise variances of the individual nodes, some papers have suggested adaptive combination methods considering the different variances (8)(9)(10) , but they still carry out suboptimal performance with different step sizes or highly correlated input signals. Recently, combination methods based on mean-square deviation (MSD) of each node were introduced (4,(11)(12) and it improved the convergence rate and steady-state error compared to other combiners.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

A Variable Step-size Diffusion Affine Projection Algorithm Using a Novel Combiner with No Need of Communicational Complexity

Seo

Kim

Koo

et al. 2016

Proceedings of the 4th IIAE International Conference on Intelligent Systems and Image Processing 2016

View full text Add to dashboard Cite

We propose a combination method for a diffusion affine projection algorithm. There are two steps to estimate unknown optimal weight in Adapt-then-combine strategy of diffusion adaptive filtering. In an adaptation step, variable step-size affine projection algorithm works to get an improvement of convergence rate and small steady-state error. In a diffusion step, the proposed combination method is applied to achieve small steady-state errors without additional communicational complexity. The cooperative strategy using information across the network achieves fast convergence rate, small steady-state error and robustness against outliers compared to a non-cooperative way. The proposed algorithm is motivated from the optimal combination rule and it has a physical meaning of weighting on how close the estimates reach in steady state.

show abstract

Novel Combination Policy for Diffusion Adaptive Networks

Fan

Luo

et al. 2019

Lecture Notes in Electrical Engineering

View full text Add to dashboard Cite

Optimal combination rules for adaptation and learning over networks

Cited by 51 publications

References 8 publications

Combination weights for diffusion strategies with imperfect information exchange

Combination weights for diffusion strategies with imperfect information exchange

A Variable Step-size Diffusion Affine Projection Algorithm Using a Novel Combiner with No Need of Communicational Complexity

Novel Combination Policy for Diffusion Adaptive Networks

Contact Info

Product

Resources

About