Factorization machine

A Factorization Machine (FM) is a general-purpose supervised learning algorithm that can be used for classification, regression, and ranking tasks.

Some details

Factorization Machines serve as a generalization of both Linear Regression and Matrix Factorization.

• Modeling Variable Interactions: While linear models only account for the effect of individual features, FMs can model interactions between features (e.g., the combination of “User Age” and “Ad Type”). This is achieved by introducing higher-order feature interactions using latent factorization.

Useful resources

• https://d2l.ai/chapter_recommender-systems/fm.html
• https://ethen8181.github.io/machine-learning/recsys/factorization_machine/factorization_machine.html
• https://docs.aws.amazon.com/sagemaker/latest/dg/fact-machines-howitworks.html
• https://cseweb.ucsd.edu/classes/fa17/cse291-b/reading/Rendle2010FM.pdf
• https://sijunhe.github.io/2018/04/19/factorization-machines.html

Tags: Machine learning Supervised machine learning

Leave a Comment