%0 Journal Article %1 vanvu %A Vu, Van %D 2010 %K perturbation singular_value_decomposition %T Singular vectors under random perturbation %U http://arxiv.org/abs/1004.2000 %X Computing the first few singular vectors of a large matrix is a problem that frequently comes up in statistics and numerical analysis. Given the presence of noise, exact calculation is hard to achieve, and the following problem is of importance: How much a small perturbation to the matrix changes the singular vectors ? Answering this question, classical theorems, such as those of Davis-Kahan and Wedin, give tight estimates for the worst-case scenario. In this paper, we show that if the perturbation (noise) is random and our matrix has low rank, then better estimates can be obtained. Our method relies on high dimensional geometry and is different from those used an earlier papers.

@article{vanvu, abstract = {Computing the first few singular vectors of a large matrix is a problem that frequently comes up in statistics and numerical analysis. Given the presence of noise, exact calculation is hard to achieve, and the following problem is of importance: How much a small perturbation to the matrix changes the singular vectors ? Answering this question, classical theorems, such as those of Davis-Kahan and Wedin, give tight estimates for the worst-case scenario. In this paper, we show that if the perturbation (noise) is random and our matrix has low rank, then better estimates can be obtained. Our method relies on high dimensional geometry and is different from those used an earlier papers.}, added-at = {2011-01-15T22:06:25.000+0100}, author = {Vu, Van}, biburl = {https://www.bibsonomy.org/bibtex/2385190fccf081e095023966d5086121d/ytyoun}, interhash = {37a216ddd1f4409dbc65c1905ca29621}, intrahash = {385190fccf081e095023966d5086121d}, keywords = {perturbation singular_value_decomposition}, timestamp = {2011-03-31T06:42:51.000+0200}, title = {Singular vectors under random perturbation}, url = {http://arxiv.org/abs/1004.2000}, year = 2010 }