Deep Recursive Embedding for High-Dimensional Data
Project description
Deep Recursive Embedding
Deep Recursive Embedding (DRE) is a novel demensionality reduction method based on a generic deep embedding network (DEN) framework, which is able to learn a parametric mapping from high-dimensional space to low-dimensional space, guided by a recursive training strategy. DRE makes use of the latent data representations for boosted embedding performance.
Lab github DRE page: Tao Lab
Maintainer's github DRE page: Xinrui Zu
MNIST embedding result
Installation
DRE can be installed with a simple PyPi command:
pip install DRE
The pre-requests of DRE are:
numpy >= 1.19
scikit-learn >= 0.16
matplotlib
numba >= 0.34
torch >= 1.0
How to use DRE
DRE follows the form of Scikit-learn APIs, whose fit_transform function is for returning the embedding result and fit for the whole model:
from DRE import DeepRecursiveEmbedding
dre = DeepRecursiveEmbedding()
# return the embedding result:
y = dre.fit_transform(x)
# or return the whole model:
dre.fit(x)
Copy and run test_mnist.py or test_mnist.ipynb to check the embedding procedure of MNIST dataset.
Release history Release notifications | RSS feed
Download files
Download the file for your platform. If you're not sure which to choose, learn more about installing packages.
Source Distribution
Built Distribution
Hashes for DRE-1.1.27-cp37-cp37m-manylinux_2_5_x86_64.manylinux1_x86_64.whl
| Algorithm | Hash digest | |
|---|---|---|
| SHA256 | d1bb99a578cfcba61f8363b879ee4244d0ad84086c6824eba14c13b563712375 |
|
| MD5 | 2f1584afdd0e87887a093ab8692a792e |
|
| BLAKE2b-256 | 0f749877d85f1df8d7fdd59818dd8e5cadc03450e7e08ed19820949451d70ba6 |
