Abstract

There are a LOT of algorithms for recommender systems. In this post, I will try to list a part of them (and hopefully a large part). I will not introduce what are recommender system, what they are for, etc ... I have a post that introduce the concept a little, but the goal here is to go straight to the point. Thus I will present each algorithm in few lines in order to just to have the big picture of each algorithm is solving and what they model. In a second part of this post, I'll try to compared them on different datasets and give the code on my github, but this will be a very painful work ! I'll try to have a dozen of algorithm codes before moving to experiments parts.
There are :
  • 11 algorithms named so far
  • 5 algorithms detailed so far
  • 0 algorithms tested experimentaly so far
  • 5 codes available

Presentation of the algorithms

Contextual Matrix Factorization - [Display] -

ALS-WR : Alternating-Least-Squares with Weighted-λ-Regularization - [Display] - [Source]

RSVD: Regularized SVD - [Display] - [Source]

imp RSVD: Improved Regularized SVD - [Display] - [Source]

FM : Factorization Machines - [Display] - [Source]

RBM for CF : Restricted Boltzmann Machine for Collaborative Filtering - [Display] - [Source]

SVD++ - [Display] - [TBD]

TimeSVD++ - [Display] - [TBD]



DFC : Divide-Factor-Combine - [Display] - [TBD]

Experiments