University Côte d'azur

ECUE Building recommender systems

ECUE's code : SMEMI315

This course belong to UE Métiers 1 MSS (3 ECTS) which contains 4 ECUE
EUR SPECTRUM
Informatique
Campus Valrose
Master 2
Semestre impair
Anglais

PRESENTATION

A recommender system (RS) can help to influence your customers’ behaviour directly but entertainingly. In this course, we will build RS’s using different approaches: content-based, collaborative filtering, context-aware, or a hybrid one. We will learn about the theory behind diverse mathematical models of an RS task: matrix and tensor decompositions, associative rules, neighbourhood methods, learning to rank, and metric learning. For the practical part, we will employ classical machine learning (such as scikit-learn), deep learning (e.g. pytorch), and a slew of specialised packages (implicit and lightfm amongs them). During the lectures, we will talk not only about theorems but also about applications of RS’s making the clients of companies and non-profit organisations happier. No prior knowledge of the subject is necessary. Python programming experience is mandatory. Statistical learning fundamentals will be nice to have.

Course's manager(s)

Boris Shminke

In class

  • 24h of lectures
  • 12h of CM
  • 12h of TD

PREREQUISITES

Before the start of the course, I must ...
  • have at least some Python programming experience

OBJECTIVES

By the end of this course, I should be able to...
  • design, build and evaluate a recommender system suitable for a particular dataset

CONTENT

Access to complete Syllabus (Authentification required)
Important
This syllabus has no contractual value. Its content is subject to change throughout this year: be aware to the last updates