Skip to main content

Kernel method for out-of-sample extension

Project description

kerneloose

Python implementation of the kernel method for out-of-sample extension (OOSE) of dimensionality reduction techniques.

Based on "Parametric nonlinear dimensionality reduction using kernel t-SNE" by Gisbrecht, Schulz, and Hammer.

The kernel method is particularly useful for projection techniques that are computationally expensive and/or have non-convex objective functions, such as t-SNE.

Installation

pip install kerneloose

Usage example

The syntax follows scikit learn conventions. Assume hd_data is a numpy array containing high-dimensional data, and an array ld_data of equal length but lower dimension was obtained by some projection technique. An OOSE of that projection can be obtained by:

from kerneloose import KernelMap

kernel_oose = KernelMap()
kernel_oose.fit(hd_data, ld_data)

The mapping can be applied to new_data (with same dimensionality as ld_data) simply by:

kernel_oose.transform(new_data)

Parameters of the calculated OOSE mapping can be saved and loaded for later use:

kernel_oose.save('some/file/name')

resume_later = KernelMap()
resume_later.load('some/file/name')

Project details


Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

kerneloose-0.0.2.tar.gz (3.6 kB view hashes)

Uploaded source

Built Distribution

kerneloose-0.0.2-py3-none-any.whl (4.8 kB view hashes)

Uploaded py3

Supported by

AWS AWS Cloud computing Datadog Datadog Monitoring Facebook / Instagram Facebook / Instagram PSF Sponsor Fastly Fastly CDN Google Google Object Storage and Download Analytics Huawei Huawei PSF Sponsor Microsoft Microsoft PSF Sponsor NVIDIA NVIDIA PSF Sponsor Pingdom Pingdom Monitoring Salesforce Salesforce PSF Sponsor Sentry Sentry Error logging StatusPage StatusPage Status page