Research project by Stanislas Bucaille under the supervision of Dr.Bonald
Abstract — In this paper, we are going to study a cutting-edge machine learning technique used for personalized recommendation: Factorization Machines (FMs). After defining what we imply by “recommendation” and “personalization”, we will see how Factorization Machines can surpass Collaborative Filtering and Content-Based methods in providing accurate understanding and predictions of consumer’s behavior using personalized variables. We will go even further and propose a modified implementation of FMs — Categorical Factorization Machines (C-FMs) — to show that there exist hidden correlations between these variables. …
