Error bound of Nyström-approximated NCut eigenvectors and its application to training size selection

@article{He2017ErrorBO,
  title={Error bound of Nystr{\"o}m-approximated NCut eigenvectors and its application to training size selection},
  author={Li He and Nilanjan Ray and Honghua Zhang},
  journal={Neurocomputing},
  year={2017},
  volume={239},
  pages={130-142}
}
We provide a novel error bound for NCut Nystrom-approximated eigenvectors.Nystrom is found identical to matrix multiplication approximation.Use sum of singular values of a specific matrix to measure the error.Give the training size bound for a user-designed tolerance of error. We provide an error bound for the popular Nystrm-approximated eigenvectors of the Normalized Cuts (NCut) out-of-sample problem. We then extend our approach to determine the size of training set given a tolerance of… CONTINUE READING

Figures, Tables, and Topics from this paper.

References

Publications referenced by this paper.