Simple methods for the optimization of complex‐valued kurtosis as a projection index. (4th March 2015)

• Record Type: Journal Article
• Authors: Hou, Siyuan; Wentzell, Peter D.; Riley, Christopher B.
• Is Part Of: Journal of chemometrics. Volume 29:Number 4(2015:Apr.)
• Page Start: 224
• Page End: 236
• Publication Date: 2015-03-04
• Subjects: Chemistry -- Mathematics -- Periodicals; Chemistry -- Statistical methods -- Periodicals; 542.85
• DOI: 10.1002/cem.2700
• Languages: English
• ISSNs: 0886-9383

Abstract

<abstract abstract-type="short"> <title> <x xml:space="preserve">Abstract</x> </title> <p>Simple and rapidly executed iterative algorithms for the optimization of the complex‐valued kurtosis as a projection index are developed, which have the following advantages: maximization and minimization in response to requirements, no requirements for sphering and strong‐uncorrelation transformation, no assumptions for circular or non‐circular source signals, and the option to choose uncorrelated scores or orthogonal projection vectors.</p> </abstract>