Influence function and robust variant of kernel canonical correlation analysis

@article{Alam2018InfluenceFA,
  title={Influence function and robust variant of kernel canonical correlation analysis},
  author={M. A. Alam and K. Fukumizu and Y. Wang},
  journal={Neurocomputing},
  year={2018},
  volume={304},
  pages={
          12-29
        }
}
  • M. A. Alam, K. Fukumizu, Y. Wang
  • Published 2018
  • Medicine, Mathematics, Computer Science
  • Neurocomputing
  • Many unsupervised kernel methods rely on the estimation of kernel covariance operator (kernel CO) or kernel cross-covariance operator (kernel CCO). Both are sensitive to contaminated data, even when bounded positive definite kernels are used. To the best of our knowledge, there are few well-founded robust kernel methods for statistical unsupervised learning. In addition, while the influence function (IF) of an estimator can characterize its robustness, asymptotic properties and standard error… CONTINUE READING

    References

    Publications referenced by this paper.
    SHOWING 1-10 OF 55 REFERENCES
    Higher-Order Regularized Kernel Canonical Correlation Analysis
    17
    Consistency and robustness of kernel-based regression in convex risk minimization
    81
    Statistical Consistency of Kernel Canonical Correlation Analysis
    204
    Hyperparameter Selection in Kernel Principal Component Analysis
    17
    A regularized kernel CCA contrast function for ICA
    26
    A Comparative Study of Kernel and Robust Canonical Correlation Analysis
    7
    Restricted kernel canonical correlation analysis
    11
    Detecting influential observations in Kernel PCA
    29