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
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
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.
Z. Zhou, X. Zu, Y. Wang, B. P. F. Lelieveldt and Q. Tao, "Deep Recursive Embedding for High-Dimensional Data," in IEEE Transactions on Visualization and Computer Graphics, vol. 28, no. 2, pp. 1237-1248, 1 Feb. 2022, doi: 10.1109/TVCG.2021.3122388.
@ARTICLE{DRE2022,
author={Zhou, Zixia and Zu, Xinrui and Wang, Yuanyuan and Lelieveldt, Boudewijn P. F. and Tao, Qian},
journal={IEEE Transactions on Visualization and Computer Graphics},
title={Deep Recursive Embedding for High-Dimensional Data},
year={2022},
volume={28},
number={2},
pages={1237-1248},
doi={10.1109/TVCG.2021.3122388}
}