Isomap
Isomap is a method for low-dimensional embedding. Isomap is used for computing a quasi-isometric, low-dimensional embedding of a set of high-dimensional data points[1].
This package defines a Isomap
type to represent a Isomap calculation results, and provides a set of methods to access its properties.
ManifoldLearning.Isomap
— TypeIsomap{NN <: AbstractNearestNeighbors} <: AbstractDimensionalityReduction
The Isomap
type represents an isometric mapping model constructed with a help of the NN
nearest neighbor algorithm.
StatsAPI.fit
— Methodfit(Isomap, data; k=12, maxoutdim=2, nntype=BruteForce)
Fit an isometric mapping model to data
.
Arguments
data
: a matrix of observations. Each column ofdata
is an observation.
Keyword arguments
k
: a number of nearest neighbors for construction of local subspace representationmaxoutdim
: a dimension of the reduced space.nntype
: a nearest neighbor construction class (derived fromAbstractNearestNeighbors
)
Examples
M = fit(Isomap, rand(3,100)) # construct Isomap model
R = predict(M) # perform dimensionality reduction
StatsAPI.predict
— Methodpredict(R::Isomap)
Transforms the data fitted to the Isomap model R
into a reduced space representation.
StatsAPI.predict
— Methodpredict(R::Isomap, X::AbstractVecOrMat)
Returns a transformed out-of-sample data X
given the Isomap model R
into a reduced space representation.
References
- 1Tenenbaum, J. B., de Silva, V. and Langford, J. C. "A Global Geometric Framework for Nonlinear Dimensionality Reduction". Science 290 (5500): 2319-2323, 22 December 2000.