Master's thesis in machine learning

This repository holds the Python and LaTeX code for my master's thesis in machine learning at UiB (INF399), with the title: "Analysis of Word Embeddings: A Clustering and Topological Approach". The master's thesis was supervised by Nello Blaser and was based on the topic from "Topology of encodings". The finalized version of the thesis can be found here.

Reposotory structure

The repository is structed as follows:

• The code directory holds all the relevant code used to perform the analysis of word embeddings, as well as some plots used in the thesis.
• The custom_figures directory holds custom figures included in the background chapter of the thesis.
• The outline directory contains the LaTeX outline of the master's thesis. Note that the information here is outdated.
• The thesis directory holds the LaTeX code used to generate the master's thesis PDF.

Running the code

First, make sure Docker is installed and up to date. Then, from the root of the repository, run the Makefile to create the code environment:

make DETACHED=1

The DETACHED=1 argument allows the Docker container to be deattached from the terminal and you can access the docker container at any time by running:

docker exec -it jtr008-docker-container /bin/bash

To enable GPUs, perform the following commands instead:

make -f Makefile-gpu DETACHED=1
docker exec -it jtr008-docker-container-gpu /bin/bash