Skip to main content
top

Bibliography

Conference Paper (international conference)

A Fast Approximate Joint Diagonalization Algorithm Using a Criterion with a Block Diagonal Weight Matrix

Tichavský Petr, Yeredor A., Nielsen Jan

: ICASSP 2008: IEEE International Conference on Acoustics, Speech, and Signal Processing, p. 3321-3324

: ICASSP 2008, IEEE International Conference on Acoustics, Speech adn Signal Processing, (Las Vegas, US, 30.03.2008-04.04.2008)

: CEZ:AV0Z10750506

: 1M0572, GA MŠk

: approximate joint diagonalization, blind source separation, autoregressive processes

(eng): We propose a new algorithm for Approximate Joint Diagonalization (AJD) with two main advantages over existing state-of-the-art algorithms: Improved overall running speed, especially in large-scale (high-dimensional) problems; and an ability to incorporate specially structured weight-matrices into the AJD criterion. The algorithm is based on approximate Gauss iterations for successive reduction of a weighted Least Squares off-diagonality criterion. The proposed Matlab implementation allows AJD of ten 100x100 matrices in 3-4 seconds (for the unweighted case) on a common PC (Pentium M, 1.86GHz, 2GB RAM), generally 3-5 times faster than the fastest competitor. The ability to incorporate weights allows fast large-scale realization of optimized versions of classical blind source separation algorithms, such as Second-Order Blind Identification (SOBI), whose weighted version (WASOBI) yields significantly improved separation performance.

(cze): V práci je navržen nový algoritmus pro vzájemnou diagonalizaci matic, který má oproti existujícím algoritmům dvě hlavní výhody: zlepšená rychlost výpočtu, zvláště u matic s vysokou dimenzí, a možnost využití specielně strukturovaných váhových matic v diagonalizačním kriteriu. Navržená implementace algoritmu v protředí Matlab umožnuje diagonalizaci 10 matic o velikosti 100 x 100 na běžném PC (Pentium M, 1.86GHz, 2GB RAM) - v průměru je 2-5x rychlejší než dosud nejrychlejší alternativy. Algoritmus umožňuje rychlou implementaci algoritmu WASOBI pro slepou separaci nezávislých autoregresních procesů s různými spektry.

: BB