An Intrinsic Dimensionality Estimator from Near-Neighbor Information

Authors:
Karl W. Pettis;Thomas A. Bailey;Anil K. Jain;Richard C. Dubes
Affiliations:
Department of Computer Science, Michigan State University, East Lansing, MI 48824.;Department of Computer Science, Michigan State University, East Lansing, MI 48824.;MEMBER, IEEE, Department of Computer Science, Michigan State University, East Lansing, MI 48824.;MEMBER, IEEE, Department of Computer Science, Michigan State University, East Lansing, MI 48824.
Venue:
IEEE Transactions on Pattern Analysis and Machine Intelligence
Year:
1979

Citing 0
Cited 7

Estimating intrinsic dimensionality of fMRI dataset incorporating an AR(1) noise model with cubic spline interpolation

Neurocomputing
Intrinsic dimension estimation of manifolds by incising balls

Pattern Recognition
Intrinsic dimension estimation by maximum likelihood in isotropic probabilistic PCA

Pattern Recognition Letters
Face detection based on the manifold

AVBPA'05 Proceedings of the 5th international conference on Audio- and Video-Based Biometric Person Authentication
Intrinsic dimension induced similarity measure for clustering

ADMA'11 Proceedings of the 7th international conference on Advanced Data Mining and Applications - Volume Part II
Dimension estimation of image manifolds by minimal cover approximation

Neurocomputing
Intrinsic dimension identification via graph-theoretic methods

Journal of Multivariate Analysis

Quantified Score

Hi-index	0.14

Visualization

Abstract

The intrinsic dimensionality of a set of patterns is important in determining an appropriate number of features for representing the data and whether a reasonable two- or three-dimensional representation of the data exists. We propose an intuitively appealing, noniterative estimator for intrinsic dimensionality which is based on nearneighbor information. We give plausible arguments supporting the consistency of this estimator. The method works well in identifying the true dimensionality for a variety of artificial data sets and is fairly insensitive to the number of samples and to the algorithmic parameters. Comparisons between this new method and the global eigenvalue approach demonstrate the utility of our estimator.