MSc in Space Technologies, Applications and Services

Machine Learning - Semester 2

Final Project on PCA and Regularized Non-Negative Matrix Factorization

This repo is my implementation for the final project of the Machine Learning course that is part of the 2nd semester of the MSc in Space Technologies, Applications and Services

This project is mainly focusing on:

• the application of Principal Components Analysis on a set of images with the aim to classify them in the season that they were taken into
• solving an optimisation problem of Regularized Non-Negative Matrix Factorization and then implementing and run it against a specific set of parameters using
  • matrices of randomly generated data
  • a real image example

It includes the following files and folders:

• machine-learning-assignment.pdf - The assignment description
• solution-Choumos.ipynb - The Jupyter Notebook with the solution
  • To see it on GitHub, just click on the Notebook and GitHub will render it
  • Or you can just clone the repo and open it through your local jupyter notebook installation
  • I will shortly add a pdf version as well
• images folder - Includes the 30 season images that are used in the PCA part of the assignment
• owl folder - Includes an image of an owl which is used at the end as an input to the Regularized NMF algorithm so that we can see the results of applying it on data that we can easily visually interpet
• README.md - This is the README file that you are currently reading.

There is no need to execute any code, you can just read the Jupyter Notebook which includes all the relevant information and describes what is being done in each step and why.

Note that the assignment description is in Greek, but the solution is in English.

In the unlikely case that you are reading this and you have any question, feel free to message me and I'll be happy to help.