Statistical Performance of Convex Tensor Decomposition
Abstract
We analyze the statistical performance of a recently proposed convex tensor decomposition algorithm. Conventionally tensor decomposition has been formulated as non-convex optimization problems, which hindered the analysis of their performance. We show under some conditions that the mean squared error of the convex method scales linearly with the quantity we call the normalized rank of the true tensor. The current analysis naturally extends the analysis of convex low-rank matrix estimation to tensors. Furthermore, we show through numerical experiments that our theory can precisely predict the scaling behaviour in practice.
Cite
Text
Tomioka et al. "Statistical Performance of Convex Tensor Decomposition." Neural Information Processing Systems, 2011.Markdown
[Tomioka et al. "Statistical Performance of Convex Tensor Decomposition." Neural Information Processing Systems, 2011.](https://mlanthology.org/neurips/2011/tomioka2011neurips-statistical/)BibTeX
@inproceedings{tomioka2011neurips-statistical,
title = {{Statistical Performance of Convex Tensor Decomposition}},
author = {Tomioka, Ryota and Suzuki, Taiji and Hayashi, Kohei and Kashima, Hisashi},
booktitle = {Neural Information Processing Systems},
year = {2011},
pages = {972-980},
url = {https://mlanthology.org/neurips/2011/tomioka2011neurips-statistical/}
}