eNet regularized BCS Framework for collaborative filtering

Blind compressive sensing with elastic net regularization for design of Recommender systems

This code solves the problem of predicting missing ratings in a recommender system by casting the problem of matrix factorization in blind compressive sensing framework with elastic net regularization.
minimize_(U,V) ||Y-A(UV)||_2 + lambda_3||U||_F + lambda_1||V||_1 + lambda_2||V||_F

Here U and V are user and item latent factor matrices.

The algorithm uses Majorization-Minimization approach to solve the above formulation in an efficient manner.

The demo file contains one test-train pair for the 100K movielens data set (available at http://grouplens.org/datasets/movielens/)