.. -- mode: rst --

|conda_version|_ |conda_downloads|_ |pypi_version|_ |pypi_downloads|_

.. |pypi_version| image:: https://img.shields.io/pypi/v/humap.svg .. _pypi_version: https://pypi.python.org/pypi/humap/

.. |pypi_downloads| image:: https://pepy.tech/badge/humap .. _pypi_downloads: https://pepy.tech/project/humap

.. |conda_version| image:: https://anaconda.org/conda-forge/humap/badges/version.svg .. _conda_version: https://anaconda.org/conda-forge/humap

.. |conda_downloads| image:: https://anaconda.org/conda-forge/humap/badges/downloads.svg .. _conda_downloads: https://anaconda.org/conda-forge/humap

.. image:: images/humap-2M.gif :alt: HUMAP exploration on Fashion MNIST dataset

===== HUMAP

Hierarchical Manifold Approximation and Projection (HUMAP) is a technique based on UMAP <https://github.com/lmcinnes/umap/>_ for hierarchical dimensionality reduction. HUMAP allows to:

1. Focus on important information while reducing the visual burden when exploring huge datasets;
2. Drill-down the hierarchy according to information demand.

The details of the algorithm can be found in our paper on ArXiv <https://arxiv.org/abs/2106.07718>_. This repository also features a C++ UMAP implementation.

Installation

HUMAP was written in C++ for performance purposes, and provides an intuitive Python interface. It depends upon common machine learning libraries, such as scikit-learn and NumPy. It also needs the pybind11 due to the interface between C++ and Python.

Requirements:

• Python 3.6 or greater
• numpy
• scipy
• scikit-learn
• pybind11
• pynndescent (for reproducible results)
• Eigen (C++)

If you have these requirements installed, use PyPI:

.. code:: bash

pip install humap

Alternatively (and preferable), you can use conda to install:

.. code:: bash

conda install humap

If using pip:

HUMAP depends on Eigen <https://eigen.tuxfamily.org/>_. Thus, make it sure to place the headers in /usr/local/include if using Unix or C:\Eigen if using Windows.

Manual installation:

For manually installing HUMAP, download the project and proceed as follows:

.. code:: bash

python setup.py bdist_wheel

.. code:: bash

pip install dist/humap*.whl

Usage examples

The simplest usage of HUMAP is as it follows:

Fitting the hierarchy

.. code:: python

import humap
from sklearn.datasets import fetch_openml


X, y = fetch_openml('mnist_784', version=1, return_X_y=True)

# build a hierarchy with three levels
hUmap = humap.HUMAP([0.2, 0.2])
hUmap.fit(X, y)

# embed level 2
embedding2 = hUmap.transform(2)

Refer to notebooks/ for complete examples.

C++ UMAP implementation

You can also fit a one-level HUMAP hierarchy, which essentially fits UMAP projection.

.. code:: python

umap_reducer = humap.UMAP()
embedding = umap_reducer.fit_transform(X)

Citation

Please, use the following reference to cite HUMAP in your work:

.. code:: bibtex

@misc{marciliojr_humap2021,
  title={HUMAP: Hierarchical Uniform Manifold Approximation and Projection}, 
  author={Wilson E. Marcílio-Jr and Danilo M. Eler and Fernando V. Paulovich and Rafael M. Martins},
  year={2021},
  eprint={2106.07718},
  archivePrefix={arXiv},
  primaryClass={cs.LG}
}

License

HUMAP follows the 3-clause BSD license and it uses the open-source NNDescent implementation from EFANNA <https://github.com/ZJULearning/efanna>. It also uses a C++ implementation of UMAP <http://github.com/lmcinnes/umap> for embedding hierarchy levels.

E-mail me (wilson_jr at outlook.com) if you like to contribute.

......