nni: Intrinsic Dimensionality Estimation from Near-Neighbor...

In ider: Various Methods for Estimating Intrinsic Dimension

Intrinsic Dimensionality Estimation from Near-Neighbor Information.

Description

nni estimates intrinsic dimension of given dataset based on the nearest-neighbor information.

Usage

nni(x, k1 = 2, k2 = 30, DM = FALSE, eps = 0.01, p = NULL)

Arguments

x	data matrix or distance matrix given by as.matrix(dist(x)).
k1	first k-NN parameter.
k2	second k-NN parameter.
DM	whether 'x' is distance matrix or not. logical.
eps	accuracy parameter.
p	ambient dimension used for automatically define 'k1' and 'k2'.

Details

First order expansion of the probability mass function is considered, then the distribution of the nearest-neighbor points from the inspection point is modeled by the Poisson distribution. The average of the nearest-distance is expressed by intrinsic dimension to be estimated.

Value

Estimated global intrinsic dimension.

Author(s)

Hideitsu Hino [e-mail]

References

B. Kegl. Intrinsic dimension estimation using packing numbers. Advances in Neural Information Processing Systems 15, 2002.

K. W. Pettis et al. An intrinsic dimensionality estimator from near neighbor information. IEEE transactions on pattern recognition and machine intelligence, 1979.

Examples

x <- gendata(DataName='SwissRoll',n=300)
estnni <- nni(x=x)
print(estnni)

ider documentation built on Feb. 16, 2023, 10:14 p.m.