M2 RAM canvas of Baptiste Malézieux for the article : Visualizing data using t-SNE written by Laurens Van der Maaten and Geoffrey Hinton

This document is a template for my report for my M2 RAM. The goal is to present the article I have read to present the contributions, but also the context and the positionning of the algorithms. In my case I have read the article of Laurens Van der Maaten and Geoffrey Hinton : ”Visualizing data using t-SNE” , published in 2008 in the Journal of Machine Learning Research.

This work is for academic purpose, the article presented is Laurens Van der Maaten and Geoffrey Hinton rights, if you find any mistakes please contact me at : baptistemalezieux@gmail.com

Introduction

The article : "Visualizing Data using t-SNE" , was written by Laurens van der Maaten and Geoffrey Hinton and was published the 11/08 in the Journal of Machine Learning Research in 2008. To introduce the t-SNE methods, today visualizing data is a used in a lot of field, but the use of high dimensional data is very complex, especially for the interpretation. That's where the authors of this article wants to improve the capability of algorithms.

A word about the authors:

  • Laurens van der Maaten is a Research Scientist at Facebook AI Research in New York, working on machine learning and computer vision. Before he worked as an Assistant Professor at Delft University of Technology, as a post-doctoral researcher at UC San Diego and as a Ph.D sutdent at Tilburg University.

  • Geoffrey Hinton is an Engineer working both for Google and University of Toronto, managing the Brain Team Toronto, which is a new part of the Google Brain Team. He has taught a free online course on Neural Networks on the education platform Coursera, and has worked at the University of Sussex, the University of California, Sand Diego and Carnegie Mellon University.

The Journal of Machine Learning Research (JMLR), present himself that it provides an international forum for the electronic and paper publication of high-quality scholarly articles in all areas of machine learning. All published papers are freely available online.

In the next parts, we will present: Context of the work, where we will explain the context of the article and talk more about the work, the difficulties and the interest about this application.
Then a Positioning part, where we are going to talk about the previous work used by the article to introduce the different approach for the visualizinsation of high dimensional data.
Then we will talk about the Contributions of the t-SNE methods but also the weakness and the strenght of this algorithms.
A part about the Experiments