Supervision: Lionel Martin

Project type: Semester project (master)


We live in a world of data but unfortunately, we still need the tools to exploit all the information that is produced. Lots of works have been done to classify entities into different classes but the complexity of these algorithms does not allow to apply them on large sets of data. Graph Signal Processing provides new tools to work with Graph objects. In particular, a faster clustering algorithm has been proposed with these methods recently. We propose to combine different techniques from GSP to enhance even further the accuracy and/or the running time of clustering algorithms.

Goal of the project:

  • Get familiar with graph signal processing and the related tools

  • Use the latest results of the field to develop a new clustering algorithm

  • Apprehend research problematic and find solutions to open problems

Student profile:

  • Familiar with coding (either Matlab or Python)

  • Eager to connect maths and computer science

  • Interested in real world data analysis (if time allows)

Project orientation:

There are two folds in the project, a theoretical approach to the problem and also an applied one. Up to the preferences of the student we can dive more into one or the other. We will probably see a bit of both.