TY - JOUR
T1 - Optimal transforms of random vectors
T2 - The case of successive optimizations
AU - Soto-Quiros, Pablo
AU - Torokhti, Anatoli
N1 - Publisher Copyright:
© 2016 Elsevier B.V.
PY - 2017/3/1
Y1 - 2017/3/1
N2 - We propose and justify new transforms of random vectors which provide, under a certain condition, better associated accuracy than that of the optimal transforms, the generic Karhunen-Loève transform and the transform considered by Brillinger. It is achieved by special structures of the proposed transforms which contain more parameters to optimize compared to the known transforms.
AB - We propose and justify new transforms of random vectors which provide, under a certain condition, better associated accuracy than that of the optimal transforms, the generic Karhunen-Loève transform and the transform considered by Brillinger. It is achieved by special structures of the proposed transforms which contain more parameters to optimize compared to the known transforms.
KW - Karhunen-Loève transform
KW - Least squares linear estimate
KW - Principal Component Analysis
KW - Rank-reduced matrix approximation
KW - Singular value decomposition
UR - http://www.scopus.com/inward/record.url?scp=84994259667&partnerID=8YFLogxK
U2 - 10.1016/j.sigpro.2016.09.020
DO - 10.1016/j.sigpro.2016.09.020
M3 - Artículo
AN - SCOPUS:84994259667
SN - 0165-1684
VL - 132
SP - 183
EP - 196
JO - Signal Processing
JF - Signal Processing
ER -