Improving the robustness to outliers of mixtures of probabilistic PCAs

Delannay, Nicolas;Archambeau, Cédric;Verleysen, Michel
(2008) 12th Pacific-Asia Conference (PAKDD 2008) — Location: Osaka (Japan) (20.May.2008)

Files

pdfdocument.pdf
  • Restricted Access
  • Adobe PDF
  • 295.21 KB
Request a copy

Details

Authors
  • Delannay, NicolasUCLouvain
    Author
  • Archambeau, CédricUCLouvain
    Author
  • Verleysen, MichelUCLouvain
    Author
Abstract
Principal component analysis, when formulated as a probabilistic model, can be made robust to outliers by using a student-t assumption on the noise distribution instead of a Gaussian one. On the other hand, mixtures of PCA is a model aimed to discover nonlinear dependencies in data by finding clusters and identifying local linear sub-manifolds. This paper shows how mixtures of PCA can be made robust to outliers too. Using a hierarchical probabilistic model, parameters are set by likelihood maximization. The method is shown to be effectively robust to outliers, even in the context of high-dimensional data.
Affiliations

Citations

  • APA
  • Chicago
  • FWB

Delannay, N., Archambeau, C., & Verleysen, M. (2008). Improving the robustness to outliers of mixtures of probabilistic PCAs. Advances in Knowledge Discovery and Data Mining, p. 527-535. https://doi.org/10.1007/978-3-540-68125-0_47