Optimal Diffusion Learning Over Networks—Part II: Multitask Algorithms

Merched, R.

doi:10.1109/ojsp.2022.3141965

Cited by 3 publications

(3 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…𝑤 ̆𝑘,𝑛 = ∑ 𝛼 𝑗,𝑘 𝜛 𝑗,𝑛 𝑗∈𝑁 𝑘 (18) Therefore, the estimate update of ATC-DQR-sRLS algorithm can be expressed as ( 19) and (20).…”

Section: The Proposed Atc-dqr-srls Algorithmmentioning

confidence: 99%

“…𝑤 ̌𝑘,𝑛 = ∑ 𝛼 𝑗,𝑘 𝜛 ̌𝑗,𝑛 𝑗∈𝑁 𝑘 (20) ATC process composes of an adaptation step in ( 17) and combination step in (18). It is noticed that the tapweighted vector can be performed in the combination step.…”

Section: The Proposed Atc-dqr-srls Algorithmmentioning

confidence: 99%

“…Sitjongsataporn [17] a class of nonlinear diffusion adaptive filtering has been furnished in the framework of orthogonal gradientbased algorithm. While, the combined diffusion affine algorithm has been presented over the distributed networks [18], Merched [19], [20] has derived with the diffusion adaptation to fuse data based on least squares mechanism against the colored inputs in the single-task and multi-task scenes. Gao et al [21] transient behavior of a multi-task diffusion on RLS has been investigated over distributed network.…”

mentioning

confidence: 99%

See 2 more Smart Citations

Diffusion recursive least squares algorithm based on triangular decomposition

Prongnuch

Sitjongsataporn

Wiangtong

2023

IJECE

View full text Add to dashboard Cite

<span lang="EN-US">In this paper, diffusion strategies used by QR-decomposition based on recursive least squares algorithm (DQR-RLS) and the sign version of DQR-RLS algorithm (DQR-sRLS) are introduced for distributed networks. In terms of the QR-decomposition method and Cholesky factorization, a modified Kalman vector is given adaptively with the help of unitary rotation that can decrease the complexity from inverse autocorrelation matrix to vector. According to the diffusion strategies, combine-then-adapt (CTA) and adapt-then-combine (ATC) based on DQR-RLS and DQR-sRLS algorithms are proposed with the combination and adaptation steps. To minimize the cost function, diffused versions of CTA-DQR-RLS, ATC-DQR-RLS, CTA-DQR-sRLS and ATC-DiQR-sRLS algorithms are compared. Simulation results depict that the proposed DQR-RLS-based and DQR-sRLS-based algorithms can clearly achieve the better performance than the standard combine-then-adapt-diffusion RLS (CTA-DRLS) and ATC-DRLS mechanisms.</span>

show abstract

“…𝑤 ̆𝑘,𝑛 = ∑ 𝛼 𝑗,𝑘 𝜛 𝑗,𝑛 𝑗∈𝑁 𝑘 (18) Therefore, the estimate update of ATC-DQR-sRLS algorithm can be expressed as ( 19) and (20).…”

Section: The Proposed Atc-dqr-srls Algorithmmentioning

confidence: 99%

Section: The Proposed Atc-dqr-srls Algorithmmentioning

confidence: 99%