vix.ing · top · new · best · stats · spec

Comparison of Clustering Methods for Extraction of Uncorrelated Sparse\n Sources from Data Mixtures

2021/08/26 by M.S. Woolfson, Woolfson, Malcolm
Chemistry · Computer Science · Medicine · #Blind Source Separation Techniques #ECG Monitoring and Analysis #Electrochemical Analysis and Applications #FOS: Electrical engineering #Signal Processing (eess.SP) #Spectroscopy and Chemometric Analyses #electronic engineering #information engineering

paper · pdf · doi:10.48550/arxiv.2108.11776

openalex publication_date 2021/08/26 · openalex created_date 2022/07/25 · openalex updated_date 2026/07/28

Abstract

There is an extensive set of methods to determine sparse sources from\nmixtures where the mixing coefficients are unknown. Each method involves\nplotting N sets of mixed data against each other in N-dimensional space. In the\napproach adopted in this paper, N dimensional normalised vectors are produced\nby joining data points that are adjacent in time. A novel clustering approach\nis adopted: the two vectors, not necessarily adjacent in time, which are\nclosest to each other are identified and one of these vectors is taken as the\nprincipal direction corresponding to one of the sources. It is shown, using a\ndeflation approach, that it is possible to estimate individual sources to\nwithin a multiplicative constant. This novel method is compared with two\nrelated methods and the standard FastICA algorithm. This new method has\ncomparable performances to three other methods when applied to examples of\npurely sparse, semi-sparse and non-sparse sources and also when applied to\nfetal ECG data.\n

Citations

Related