Fast Laplacian Eigenmaps: lightweight multicore LE for non-linear dimensional reduction with minimal memory usage
Open-source Laplacian Eigenmaps for dimensionality reduction of large data in python. Comes with an
wrapper for NMSlib to compute approximate-nearest-neighbors.
Performs several times faster than the default scikit-learn implementation.
You’ll need NMSlib for using this package properly. Installing it with no binaries is recommended if your CPU supports
advanced instructions (it problably does):
pip3 install --no-binary :all: nmslib
# Along with requirements:
pip3 install numpy pandas scipy scikit-learn
Then you can install this package with pip:
See the following example with the handwritten