A Fast Distributed Non-Negative Matrix Factorization Algorithm Based on DSGD

In numerous solutions, the distributed stochastic gradient descent algorithm is one of the most popular algorithms for parallelization matrix decomposition. However, in parallel calculation, the computing speed of each computing node was greatly different because of the imbalance of the computing no...

Celý popis

Uloženo v:
Podrobná bibliografie
Vydáno v:International journal of distributed systems and technologies Ročník 9; číslo 3; s. 24 - 38
Hlavní autoři: Chen, Baifan, Gao, Yan, Zhou, Lingjun, Xing, Xiaobing
Médium: Journal Article
Jazyk:angličtina
Vydáno: Hershey IGI Global 01.07.2018
Témata:
Algorithms
Computation
Control algorithms
Convergence
Decomposition
Efficiency
Nodes
Parallel processing
Recommender systems
Scheduling
Variables
China
ISSN:1947-3532, 1947-3540
On-line přístup:Získat plný text
Tagy: Přidat tag
Žádné tagy, Buďte první, kdo vytvoří štítek k tomuto záznamu!
Popis
Shrnutí:In numerous solutions, the distributed stochastic gradient descent algorithm is one of the most popular algorithms for parallelization matrix decomposition. However, in parallel calculation, the computing speed of each computing node was greatly different because of the imbalance of the computing nodes. This article reduced the data skew for all computing nodes during distributed execution to solve the problem of locking waiting. The improved algorithm on DSGD was named D-DSGD, which reduced the time consumption of the algorithm and improved the utilization rate of the nodes. Meanwhile, the dynamic step size adjusting strategy was applied to improve the convergence rate of the algorithm. To ensure the non-negative matrix decomposition, non-negative control was added into D-DSGD and the improved algorithm was named D-NMF. Compared with the existing methods, the proposed algorithm in this article has a marked impact on reducing the latency and speed of convergence.
Bibliografie:ObjectType-Article-1
SourceType-Scholarly Journals-1
ObjectType-Feature-2
content type line 14
ISSN:1947-3532
1947-3540
DOI:10.4018/IJDST.2018070102