ScholarLens Back to search
2011•IEEE Signal Processing LettersRequires access

Stable and Orthonormal OJA Algorithm With Low Complexity

Rong Wang, Minli Yao, Daoming Zhang, Hongxing Zou

Open publisher page 7 citations

Abstract

In this letter, a stable and orthonormal version of the OJA algorithm (SOOJA) is investigated for principal and minor subspace extraction and tracking. The new algorithm presented here guarantees the orthonormality of the weight matrix at each iteration through a novel orthonormalization method. Moreover, it obtains both a high numerical stability and a low computational complexity. The superiority of the proposed algorithm to some existing subspace tracking algorithms is demonstrated using a classical example. Simulation results confirm the veracity of the subspace tracking algorithm advocated.

About this research paper

What this paper is about

In this letter, a stable and orthonormal version of the OJA algorithm (SOOJA) is investigated for principal and minor subspace extraction and tracking. The new algorithm presented here guarantees the orthonormality of the weight matrix at each iteration through a novel orthonormalization method. Moreover, it obtains both a high numerical stability and a low computational complexity. The superiority of the proposed algorithm to some existing subspace tracking algorithms is demonstrated using a classical example. Simulation results confirm the veracity of the subspace tracking algorithm advocated.

Why it matters

OpenAlex reports 7 citations for this work. Citation counts describe recorded attention and do not establish research quality.

Key contribution

A contribution statement is not available in the OpenAlex record.

Method / approach

Method details are not available in the OpenAlex metadata.

Main findings

Findings are not separately available in the OpenAlex metadata.

Limitations

Limitations are not available in the OpenAlex metadata.

Applications

Application details are not available in the OpenAlex metadata.

Available abstract

In this letter, a stable and orthonormal version of the OJA algorithm (SOOJA) is investigated for principal and minor subspace extraction and tracking. The new algorithm presented here guarantees the orthonormality of the weight matrix at each iteration through a novel orthonormalization method. Moreover, it obtains both a high numerical stability and a low computational complexity. The superiority of the proposed algorithm to some existing subspace tracking algorithms is demonstrated using a classical example. Simulation results confirm the veracity of the subspace tracking algorithm advocated.

Key concepts: Orthonormality, Orthonormal basis, Subspace topology, Algorithm, Computational complexity theory, Computer science, Stability (learning theory), Tracking (education)

Related papers

Related paper 1Related paper 2Related paper 3
Back to paper searchBrowse research topicsOriginal source
Stable and Orthonormal OJA Algorithm With Low Complexity — Research Paper | ScholarLens