A Unified Framework for Outlier-Robust PCA-like Algorithms

Wenzhuo Yang, Huan Xu
Proceedings of the 32nd International Conference on Machine Learning, PMLR 37:484-493, 2015.

Abstract

We propose a unified framework for making a wide range of PCA-like algorithms – including the standard PCA, sparse PCA and non-negative sparse PCA, etc. – robust when facing a constant fraction of arbitrarily corrupted outliers. Our theoretic analysis establishes solid performance guarantees of the proposed framework: its estimation error is upper bounded by a term depending on the intrinsic parameters of the data model, the selected PCA-like algorithm and the fraction of outliers. Comprehensive experiments on synthetic and real-world datasets demonstrate that the outlier-robust PCA-like algorithms derived from our framework have outstanding performance.

Cite this Paper

BibTeX
@InProceedings{pmlr-v37-yangc15, title = {A Unified Framework for Outlier-Robust PCA-like Algorithms}, author = {Yang, Wenzhuo and Xu, Huan}, booktitle = {Proceedings of the 32nd International Conference on Machine Learning}, pages = {484--493}, year = {2015}, editor = {Bach, Francis and Blei, David}, volume = {37}, series = {Proceedings of Machine Learning Research}, address = {Lille, France}, month = {07--09 Jul}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v37/yangc15.pdf}, url = {https://proceedings.mlr.press/v37/yangc15.html}, abstract = {We propose a unified framework for making a wide range of PCA-like algorithms – including the standard PCA, sparse PCA and non-negative sparse PCA, etc. – robust when facing a constant fraction of arbitrarily corrupted outliers. Our theoretic analysis establishes solid performance guarantees of the proposed framework: its estimation error is upper bounded by a term depending on the intrinsic parameters of the data model, the selected PCA-like algorithm and the fraction of outliers. Comprehensive experiments on synthetic and real-world datasets demonstrate that the outlier-robust PCA-like algorithms derived from our framework have outstanding performance.} }
Endnote
%0 Conference Paper %T A Unified Framework for Outlier-Robust PCA-like Algorithms %A Wenzhuo Yang %A Huan Xu %B Proceedings of the 32nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2015 %E Francis Bach %E David Blei %F pmlr-v37-yangc15 %I PMLR %P 484--493 %U https://proceedings.mlr.press/v37/yangc15.html %V 37 %X We propose a unified framework for making a wide range of PCA-like algorithms – including the standard PCA, sparse PCA and non-negative sparse PCA, etc. – robust when facing a constant fraction of arbitrarily corrupted outliers. Our theoretic analysis establishes solid performance guarantees of the proposed framework: its estimation error is upper bounded by a term depending on the intrinsic parameters of the data model, the selected PCA-like algorithm and the fraction of outliers. Comprehensive experiments on synthetic and real-world datasets demonstrate that the outlier-robust PCA-like algorithms derived from our framework have outstanding performance.
RIS
TY - CPAPER TI - A Unified Framework for Outlier-Robust PCA-like Algorithms AU - Wenzhuo Yang AU - Huan Xu BT - Proceedings of the 32nd International Conference on Machine Learning DA - 2015/06/01 ED - Francis Bach ED - David Blei ID - pmlr-v37-yangc15 PB - PMLR DP - Proceedings of Machine Learning Research VL - 37 SP - 484 EP - 493 L1 - http://proceedings.mlr.press/v37/yangc15.pdf UR - https://proceedings.mlr.press/v37/yangc15.html AB - We propose a unified framework for making a wide range of PCA-like algorithms – including the standard PCA, sparse PCA and non-negative sparse PCA, etc. – robust when facing a constant fraction of arbitrarily corrupted outliers. Our theoretic analysis establishes solid performance guarantees of the proposed framework: its estimation error is upper bounded by a term depending on the intrinsic parameters of the data model, the selected PCA-like algorithm and the fraction of outliers. Comprehensive experiments on synthetic and real-world datasets demonstrate that the outlier-robust PCA-like algorithms derived from our framework have outstanding performance. ER -
APA
Yang, W. & Xu, H.. (2015). A Unified Framework for Outlier-Robust PCA-like Algorithms. Proceedings of the 32nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 37:484-493 Available from https://proceedings.mlr.press/v37/yangc15.html.

Related Material