Improving regularized singular value decomposition for collaborative filtering

By Arkadiusz Paterek. 2007

909 citations recorded by the source corpus.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Variational Autoencoders for Collaborative Filtering (2018, background): “Asymmetric matrix factorization [35] may also be interpreted as an autoencoder, as elaborated in Steck [44].”
Gaussian Ranking by Matrix Factorization (2015, methodology): “…of positive and negative items of a given user u is denoted by Inthispaper, we use the asymmetric matrix factorization model (AMF) introduced by [9] and later extended to SVD++ in [6]: the latent user-vector is deﬁned as the sum of the latent item-vectors (cid:126)q j of the positive items j of…”
Guaranteed Matrix Completion via Non-Convex Factorization (2014, background): “Problem (4) is a non-convex fourth-order polynomial optimization problem, and can be solved to stationary points by standard nonlinear optimization algorithms such as gradient descent method, alternating minimization [1], [18], [19], [21] and SGD (stochastic gradient descent) [1], [16], [23], [24].”
An Efficient Non-Negative Matrix-Factorization-Based Approach to Collaborative Filtering for Recommender Systems (2014, methodology): “The representative works include biased SVD model [14], biased regularized incremental simultaneous MF model [15], model [3], [16], probabilistic MF model [17], and nonparametric MF model [18].”
Adaptive bootstrapping of recommender systems using decision trees (2011, background): “A common practice is assuming an additive parameterization of the bias, where bui = bu+bi [11, 15].”
Circle-based recommendation in online social networks (2012, methodology): “[5, 7, 8, 20, 21]), we here briefly review the basic low-rank matrix factorization (MF) approach, which will be extended towards social network information in the remainder of this paper.”
Item popularity and recommendation accuracy (2011, background): “This is interesting, as some of these approaches, like [16, 19, 13, 20], actually account in some sense for the fact that ratings are missing not at random from the observed data–but in the context of observed ratings (as to minimize RMSE).”
SLIM: Sparse Linear Methods for Top-N Recommender Systems (2011, background): “Paterek [15] introduced a linear model for each item for rating prediction, in which the rating of a user ui on an item tj is calculated as the aggregation of the ratings of ui on all other items.”
A Distributed, Architecture-Centric Approach to Computing Accurate Recommendations from Very Large and Sparse Datasets (2011, background): “in a single RSVD estimate with over 96 features [121].”
Functional matrix factorizations for cold-start recommendation (2011, methodology): “Recently, matrix factorization becomes a popular direction for collaborative ﬁltering [12,14,16,21,24].”
A survey of collaborative filtering based social recommender systems (2014, methodology): “One of the most accurate approaches was found to be matrix factorization [24,5,6,13].”
Learning Attribute-to-Feature Mappings for Cold-Start Recommendations (2010, methodology): “Pilaszy and Tikk [21] propose an MF model for rating prediction that maps attributes to the factor space using a linear transformation, based on a method proposed by Paterek [19].”
A Coupled Indian Buffet Process Model for Collaborative Filtering (2012, methodology): “…and significance weighting (nearest-neighbor) (Breese et al., 1998), a regularized gradient descent SVD method with user and item biases (Paterek, 2007), and the model-based BLITR method (Harvey et al., 2011), which postulates a statistical model inspired by principles employed in LDA.”
Temporal Pattern-Aware QoS Prediction via Biased Non-Negative Latent Factorization of Tensors (2020, methodology): “It is widely accepted that LBs can enlarge the solution space of a resulting model [31]–[33].”