https://github.com/maayanlab/graph-dr
Benchmarks of dimensionality reduction algorithms
https://github.com/maayanlab/graph-dr
autoencoder cytoscape dimensionality-reduction force-layout graph-algorithms graph-theory manifold-learning
Last synced: 5 months ago
JSON representation
Benchmarks of dimensionality reduction algorithms
- Host: GitHub
- URL: https://github.com/maayanlab/graph-dr
- Owner: MaayanLab
- Created: 2018-03-06T16:26:30.000Z (about 7 years ago)
- Default Branch: master
- Last Pushed: 2023-03-24T21:55:05.000Z (about 2 years ago)
- Last Synced: 2023-03-31T16:29:38.583Z (about 2 years ago)
- Topics: autoencoder, cytoscape, dimensionality-reduction, force-layout, graph-algorithms, graph-theory, manifold-learning
- Language: Jupyter Notebook
- Size: 31.1 MB
- Stars: 2
- Watchers: 4
- Forks: 3
- Open Issues: 2
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
Graph-DR: Benchmarking of dimensionality reduction algorithms
==============
A series of Jupyter Notebooks experimenting on different dimensionality reduction algorithms and their performance on the [MNIST dataset](http://yann.lecun.com/exdb/mnist/).
Get started
------------
There are two options to run the notebooks in this repository:
### 1. Docker
To start running the docker image:
docker run -it -p 8888:8888 maayanlab/graph-dr:2019
Next, you can open a browser and go to http://localhost:8888. You will be required to enter the token to access the notebook server, which can be found in the terminal running the Docker image.
### 2. virtualenv
The recommended way to run these notebooks live is to set up a isolated Python envrionment using [virtualenv](https://virtualenv.pypa.io/en/stable/), after cloning this repository:
git clone https://github.com/MaayanLab/Graph-DR.git
cd Graph-DR/
Run the following to set up a Python virtural environment:
virtualenv venv
Then activate the virtural environment and install the required Python packages:
source venv/bin/activate
pip install -r requirements.txt
Next, you can start a Jupyter server:
jupyter notebook
Other dependencies
------------
Some code blocks in the notebooks require [Cytoscape](http://www.cytoscape.org/) (>3.5.1) to be running as the background.
To run the firework layout, you will also need to install the a Cytoscape app [AllegroLayout](https://www.dropbox.com/s/uwcpjes6vv212fm/allegrolayout-2.2.2.jar?dl=0). Once the jar file is downloaded, go to `Apps` -> `Install from File` -> open the jar file.
References
------------
+ [Making sense of principal component analysis, eigenvectors & eigenvalues](https://stats.stackexchange.com/questions/2691/making-sense-of-principal-component-analysis-eigenvectors-eigenvalues)
+ [MNIST For ML Beginners](https://www.tensorflow.org/get_started/mnist/beginners)
+ [Visualizing MNIST: An Exploration of Dimensionality Reduction](http://colah.github.io/posts/2014-10-Visualizing-MNIST/)
+ [Tensorflow Embedding Projector](http://projector.tensorflow.org/)
+ [van der Maaten's t-SNE page](http://lvdmaaten.github.io/tsne/)
+ [van der Maaten et al.: Dimensionality Reduction: A Comparative Review](https://www.tilburguniversity.edu/upload/59afb3b8-21a5-4c78-8eb3-6510597382db_TR2009005.pdf)
+ [van der Maaten: Learning a Parametric Embedding by Preserving Local Structure](http://lvdmaaten.github.io/publications/papers/AISTATS_2009.pdf)
+ [Kokiopoulou and Saad: Enhanced graph-based dimensionality reduction with repulsion Laplaceans](http://www.sciencedirect.com/science/article/pii/S0031320309001460)
+ [UMAP: Uniform Manifold Approximation and Projection for Dimension Reduction](https://arxiv.org/abs/1802.03426)
+ [Unsupervised learning: the curious pupil](https://deepmind.com/blog/unsupervised-learning/)