OECD R&I conference | Tutorial on word2vec, t-SNE and Gaussian Mixtures

I presented this tutorial at an OECD Research and Innovation conference that was co-organised with Nesta. I focused on a practical application, namely how to cluster UK research projects to thematic topics using text data.

I went through methods such as word2vec to create document vectors, t-SNE for dimensionality reduction and Gaussian Mixtures for fuzzy clustering.

Link to Jupyter Notebook