In this section, we will see two examples of developing a more robust recommendation system using FMs. We will start with a brief explanation of FM and their application to the cold-start recommendation problem.
Then we will see a short example of using an FM to developing a real-life recommendation system. After that, we will see an example using an improved version of the FM algorithm called a Neural Factorization Machine (NFM).
FM-based techniques are at the cutting edge of personalization. They have proven to be extremely powerful with enough expressive capacity to generalize existing models, such as matrix/tensor factorization and polynomial kernel regression. In other words, this type of algorithm is a supervised learning approach that enhances the performance of linear models by incorporating second-order feature interactions that are absent in matrix factorization algorithms.
Existing recommendation algorithms require...