Low Rank Approximation Techniques Project ( ❛ ͜ʖ ❛ )

A project for the course Low-rank Approximation Techniques.

+ =

Code:

• The notebook main.ipynb contain all runnable code and output all the figures. Change the number of the images in the main to chose which original images you want to use.
• The python file sinkhorn.py contain the main function Sinkhorn for the sinkhorn algorithm and variations.
• The python file builders.py contain the builders functions used to manipulate images and initialize parameters of Sinkhorn.

Dependencies:

• Numpy
• Matplotlib
• Sikit-Learn

Main references:

• [1] Massively scalable Sinkhorn distances via the Nyström method
• [2] Computational Optimal Transport
• [3] Adaptive color transfer with relaxed optimal transport