On GPU–CUDA as preprocessing of fuzzy-rough data reduction by means of singular value decomposition
- 103 Downloads
Data reduction algorithms often produce inaccurate results for loss of relevant information. Recently, the singular value decomposition (SVD) method has been used as preprocessing method in order to deal with high-dimensional data and achieve fuzzy-rough reduct convergence on higher dimensional datasets. Despite the well-known fact that SVD offers attractive properties, its high computational cost remains a critical issue. In this work, we present a parallel implementation of the SVD algorithm on graphics processing units using CUDA programming model. Our approach is based on an iterative parallel version of the QR factorization by means of Givens rotations using the Sameh and Kuck scheme. Our results show significant improvements in terms of performances with respect to the CPU version that encourage its usability for this expensive processing of data.
KeywordsSVD algorithm GPU computing Performance evaluation
Compliance with ethical standards
This paper is not funded by a project.
Conflict of interest
All authors declare the absence of conflicts of interest.
This article does not contain any studies with human participants or animals performed by any of the authors.
- Cuomo S, Galletti A, Marcellino L (2015) A GPU algorithm in a distributed computing system for 3d MRI denoising. In: Proceedings—2015 10th international conference on P2P, parallel, grid, cloud and internet computing, 3PGCIC 2015, pp 557–562Google Scholar
- Cuomo S, De Michele P, Galletti A, Marcellino L (2016) A GPU parallel implementation of the local principal component analysis overcomplete method for DW image denoising. In: Proceedings—IEEE symposium on computers and communications. pp 26-31. doi: 10.1109/ISCC.2016.7543709
- Cuomo S, De Michele P, Galletti A, Marcellino L (2016) A parallel PDE-based numerical algorithm for computing the optical flow in hybrid systems. J Comput Sci. doi: 10.1016/j.jocs.2017.03.011. Article in Press
- Cuomo S, De Michele P, Galletti A, Marcellino L (2016) Local principal component analysis overcomplete method: a GPU parallel implementation combining shared and global memories. In: International conference on high performance computing and simulation, HPCS, pp 81–87. doi: 10.1109/HPCSim.2016.7568319
- Cuomo S, De Michele P, Maiorano F, Marcellino L (2016) Gpu profiling of singular value decomposition in olpca method for image denoising. In: International conference on P2P. Cloud and internet computing. Springer, Parallel, Grid, pp 707–716Google Scholar
- D’Amore L, Marcellino L, Mele V, Romano D (2012) Deconvolution of 3D fluorescence microscopy images using graphics processing units. In: Lecture notes in computer science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes. Bioinformatics vol. 7203, no. 1, pp 690–699Google Scholar
- Pudil P, Novoviov J (1998) Novel methods for feature subset selection with respect to problem knowledge. In: Liu H, Motoda H (eds). Feature extraction, construction and selection. p. 101. doi: 10.1007/978-1-4615-5725-8-7. ISBN 978-1-4613-7622-4
- Rama Devi Y, Venu Gopal P, Sai Prasad P (2011) Fuzzy rough data reduction using SVD. Int J Comput Electr Eng 3(3):384–388Google Scholar
- Richard J, Shen Q (2002) Fuzzy-rough sets for descriptive dimensionality reduction. In: Fuzzy systems, proceedings of the 2002 IEEE international conference, pp 29–34Google Scholar