Regularized Singular Value Decomposition and Application to Recommender System
This work addresses the need for more robust matrix decomposition techniques in machine learning applications like recommender systems, though it appears incremental.
The paper tackles the problem of improving singular value decomposition (SVD) by introducing regularization, resulting in a method that significantly outperforms SVD in recommender system experiments.
Singular value decomposition (SVD) is the mathematical basis of principal component analysis (PCA). Together, SVD and PCA are one of the most widely used mathematical formalism/decomposition in machine learning, data mining, pattern recognition, artificial intelligence, computer vision, signal processing, etc. In recent applications, regularization becomes an increasing trend. In this paper, we present a regularized SVD (RSVD), present an efficient computational algorithm, and provide several theoretical analysis. We show that although RSVD is non-convex, it has a closed-form global optimal solution. Finally, we apply RSVD to the application of recommender system and experimental result show that RSVD outperforms SVD significantly.