Crate numcodecs_random_projection

Expand description

Random Projection codec implementation for the numcodecs API.

Structs§

ClosedOpenUnit: Floating point number in $[0.0, 1.0)$
OpenClosedUnit: Floating point number in $(0.0, 1.0]$
RandomProjectionCodec: Codec that uses random projections to reduce the dimensionality of high- dimensional data to compress it.

RandomProjectionCodecError: Errors that may occur when applying the RandomProjectionCodec.
RandomProjectionKind: Projection kind that is used to generate the random projection matrix
RandomProjectionReduction: Method with which the reduced dimensionality $K$ is selected

density_or_ping_li_minimum: Extract the provided density if it is Some(_), or compute the minimum required density $\frac{1}{\sqrt{d}}$ as recommended by Li et al .
johnson_lindenstrauss_min_k: Find a ‘safe’ number of components $K$ to randomly project to.
project_into: Applies random projection to the input data and outputs into the projected array.
project_with_projection: Applies random projection to the input data with the given seed, reduction method, and projection kind and returns the resulting projected array.
reconstruct: Applies the (approximate) inverse of random projection to the projected array to reconstruct the input data with dimensionality d and returns the resulting reconstructed array.
reconstruct_into: Applies the (approximate) inverse of random projection to the projected array to reconstruct the input data outputs into the reconstructed array.
reconstruct_into_with_projection: Applies the (approximate) inverse of random projection to the projected array to reconstruct the input data with the given seed and projection kind and outputs into the reconstructed array.
reconstruct_with_projection: Applies the (approximate) inverse of random projection to the projected array.