Bayesian learning for robust principal component analysis
2015 (English)In: 2015 23rd European Signal Processing Conference, EUSIPCO 2015, IEEE , 2015, 2361-2365 p.Conference paper (Refereed)
We develop a Bayesian learning method for robust principal component analysis where the main task is to estimate a low-rank matrix from noisy and outlier contaminated measurements. To promote low-rank, we use a structured Gaussian prior that induces correlations among column vectors as well as row vectors of the matrix under estimation. In our method, the noise and outliers are modeled by a combined noise model. The method is evaluated and compared to other methods using synthetic data as well as data from the MovieLens 100K dataset. Comparisons show that the method empirically provides a significant performance improvement over existing methods.
Place, publisher, year, edition, pages
IEEE , 2015. 2361-2365 p.
Robust principal component analysis, matrix completion, Bayesian learning
Research subject Electrical Engineering
IdentifiersURN: urn:nbn:se:kth:diva-185726DOI: 10.1109/EUSIPCO.2015.7362807ISI: 000377943800474ScopusID: 2-s2.0-84963956159OAI: oai:DiVA.org:kth-185726DiVA: diva2:923044
23rd European Signal Processing Conference, EUSIPCO 2015; Nice Congress CenterNice; France; 31 August 2015 through 4 September 2015
FunderSwedish Research Council, 621-2011-5847
QC 201604262016-04-252016-04-252016-07-15Bibliographically approved