Objective Empirical Mode Decomposition metric
| Autor: | O. Cadenas, Slawomir J. Nasuto, Dawid Laszuk |
|---|---|
| Rok vydání: | 2016 |
| Předmět: |
business.industry
Pattern recognition Machine learning computer.software_genre Instantaneous phase Hilbert–Huang transform lcsh:Telecommunication Visual inspection Set (abstract data type) symbols.namesake Amplitude lcsh:TK5101-6720 Component (UML) Metric (mathematics) symbols Hilbert transform Artificial intelligence Data mining business computer Mathematics |
| Zdroj: | TSP International Journal of Advances in Telecommunications, Electrotechnics, Signals and Systems, Vol 5, Iss 1, Pp 8-14 (2016) |
| ISSN: | 1805-5443 |
| DOI: | 10.11601/ijates.v5i1.139 |
| Popis: | Empirical Mode Decomposition (EMD) is a data-driven technique for extraction of oscillatory components from data. Although it has been introduced over 15 years ago, its mathematical foundations are still missing which also implies lack of objective metrics for decomposed set evaluation. Currently, the most common technique for assessing results of EMD is their visual inspection, which is very subjective. This article provides objective measures for validating EMD output, which are derived from the original definition of oscillatory components. Three proposed metrics refer to component's idealised characteristics, i.e. its significant instantaneous frequency and the ability to extract amplitude- and frequency-modulated parts. Possible application of these metrics is presented on two examples. |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|