Title :
The equivalence of TLS and correspondence analysis
Author :
Yao, I. ; Lorenzelli, F. ; Kong, J.
Author_Institution :
Dept. of Eng., California Univ., Los Angeles, CA, USA
Abstract :
The total least squares (TLS) technique is will known in numerical analysis and modern signal processing, while correspondence analysis (CA) has found applications in applied data and clustering analysis. Both techniques use the SVD method to determine their respective optimum solutions. Upon an appropriate preprocessing operation of centering the data matrix, it is shown by minimizing the energy of the perturbation imposed on the data matrix that the two approaches are equivalent. Some simple graphical interpretations and examples are given
Keywords :
least squares approximations; matrix algebra; numerical analysis; signal processing; TLS; clustering analysis; correspondence analysis; data analysis; data matrix; graphical interpretations; numerical analysis; perturbation energy minimization; preprocessing operation; signal processing; total least squares; Curve fitting; Data analysis; Equations; Frequency estimation; Gaussian processes; Least squares approximation; Least squares methods; Numerical analysis; Signal analysis; Signal processing;
Conference_Titel :
Acoustics, Speech, and Signal Processing, 1992. ICASSP-92., 1992 IEEE International Conference on
Conference_Location :
San Francisco, CA
Print_ISBN :
0-7803-0532-9
DOI :
10.1109/ICASSP.1992.226622