1D Multivariate Empirical Mode Decomposition (MEMD) | Part 1

Introduction to the 1D multivariate empirical mode decomposition (MEMD).

The video explains why the MEMD should be used to process multivariate data, i.e., multiple 1D signals, rather than the univariate EMD and exemplarily shows how well frequencies/scales are aligned across multiple signals. It provides a high-level overview of the MEMD and how the algorithm distinguishes from the univariate version. Special attention is given to the signal’s projection, which is a key step in performing the MEMD. A detailed discussion of each step involved in the algorithm will be given in the next video (see link below).

univariate EMD:    • Empirical Mode Decomposition (1D, univaria...  
MEMD, Part 2:    • 1D Multivariate Empirical Mode Decompositi...  

----------------------------------------------------------------------------------------------------------------------------------------------------

References:
Rehman, N., & Mandic, D. P. (2010). Multivariate empirical mode decomposition. Proceedings of the Royal Society A: Mathematical, Physical and Engineering Sciences, 466(2117), 1291-1302. https://doi.org/10.1098/rspa.2009.0502


Personal webpage of Prof. Mandic, where you can get the matlab code: https://www.commsp.ee.ic.ac.uk/~mandi...

----------------------------------------------------------------------------------------------------------------------------------------------------

Time schedule:
00:00 Introduction
01:50 Most important property of the MEMD
04:10 Exemplary application to synthetic data
09:58 Recap of the univariate EMD
12:12 Differences of multivariate algorithm
14:32 Signal projections