EMD performance comparison: single vs double floating points

Laszuk, Dawid and Cadenas, Jose O. and Nasuto, Slawomir J. (2016) EMD performance comparison: single vs double floating points. International journal of signal processing systems, 4 (4). pp. 349-353. ISSN 2315-4535. (doi:https://doi.org/10.18178/ijsps.4.4.349-353) (Full text available)

PDF - Author's Accepted Manuscript
Download (887kB) Preview
[img]
Preview
Official URL
http://www.ijsps.com/uploadfile/2016/0831/20160831...

Abstract

Empirical mode decomposition (EMD) is a data-driven method used to decompose data into oscillatory components. This paper examines to what extent the defined algorithm for EMD might be susceptible to data format. Two key issues with EMD are its stability and computational speed. This paper shows that for a given signal there is no significant difference between results obtained with single (binary32) and double (binary64) floating points precision. This implies that there is no benefit in increasing floating point precision when performing EMD on devices optimised for single floating point format, such as graphical processing units (GPUs).

Item Type: Article
Uncontrolled keywords: Terms—Empirical Mode Decomposition, Floating Point Arithmetic, Intrinsic Mode Function, Performance Test, Signal Decomposition1
Subjects: Q Science > QA Mathematics (inc Computing science) > QA 76 Software, computer programming, > QA76.76 Computer software
Q Science > QA Mathematics (inc Computing science) > QA 76 Software, computer programming, > QA76.87 Neural computers, neural networks
Divisions: Faculties > Sciences > School of Engineering and Digital Arts
Faculties > Sciences > School of Engineering and Digital Arts > Instrumentation, Control and Embedded Systems
Depositing User: Jose Oswaldo Cadenas
Date Deposited: 05 Oct 2016 08:40 UTC
Last Modified: 03 Feb 2017 10:01 UTC
Resource URI: https://kar.kent.ac.uk/id/eprint/57350 (The current URI for this page, for reference purposes)
Cadenas, Jose O.: https://orcid.org/0000-0003-4152-6458
  • Depositors only (login required):

Downloads

Downloads per month over past year