COFI RANK - Maximum Margin Matrix Factorization for Collaborative Ranking

By Markus Weimer, Alexandros Karatzoglou, Quoc V. Le, Alexander J. Smola. 2007

229 citations recorded by the source corpus.

Subfields: Recommender Systems, Statistical Learning Theory

How researchers see this paper

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

In Recommender Systems

In Statistical Learning Theory

Citations in the corpus

Variational Autoencoders for Collaborative Filtering (2018, background): “Top-N ranking loss is difficult to optimize directly and previous work on direct ranking loss minimization resorts to relaxations and approximations [49, 50].”
Recurrent Neural Networks with Top-k Gains for Session-based Recommendations (2017, background): “One of the first learning to rank techniques for collaborative filtering was introduced in [20].”
Gaussian Ranking by Matrix Factorization (2015, background): “There has been a large body of work on training collaborative ﬁltering approaches as to maximize various ranking metrics, see eg [5, 8, 17, 2, 18, 10, 20, 21, 14, 13, 12].”
Recommendations as Treatments: Debiasing Learning and Evaluation (2016, methodology): “bjective for other u;i(Y;Y^) that are nondifferentiable is more challenging, but possible avenues exist (Joachims,2005;Chapelle & Wu,2010). Finally, note that other recommendation methods (e.g.,Weimer et al., 2007;Lin,2007) can in principle be adapted to propensity scoring as well. 5. Propensit”
Slow Learners are Fast (2009, background): “This occurs, for instance, in the case of conditional random fields [15, 18], in planning [14], and in ranking [19].”
Proceedings of the Twenty-Third International Joint Conference on Artificial Intelligence GBPR: Group Preference Based Bayesian Personalized Ranking for One-Class Collaborative Filtering (unknown): “…feedback of “1” and “2”, where some regression or ranking loss functions were designed to ﬁt the score or to preserve the ordering, among which matrix-factorization based models have been shown to be the most effective solutions [Weimer et al. , 2007; Koren, 2008; Rendle, 2012; Liu et al. , 2013].”
Improving maximum margin matrix factorization (2008, methodology): “The extensions have been introduced in a way that preserves recent extensions of MMMF to structured loss (Weimer et al. 2008).”
An accelerated gradient method for trace norm minimization (2009, methodology): “Srebro, N., Rennie, J. D. M., & Jaakkola, T. S. (2005).”
Interactive collaborative filtering (2013, unknown): “In addition, we add a constraint for all the algorithms that the same item should not be repeatedly recommended as suggested in most previous ranking-oriented recommendation settings [28, 40].”
Optimizing top-n collaborative filtering via dynamic negative item sampling (2013, background): “This is essentially different from most ranking-oriented CF algorithms [6, 7, 9] where firstly the target ranking function is somehow smoothed and then a pair-wise or list-wise learning method is performed.”
Training and testing of recommender systems on data missing not at random (2010, methodology): “In support of the validity to use observed ratings only, the experiments on the Net(cid:13)ix data in [8] show that models with better RMSE on the observed ratings achieve a better top-k hit rate on all items.”
Recommender Systems (2016, unknown): “The works in [180, 500, 569, 624] explore maximum margin factorization, which is very closely related to unconstrained matrix factorization.”
Tighter Bounds for Structured Estimation (2008, background): “This could be, for instance, the number of bits in which y and y differ, i.e. ∆(y, y ) = y − y 1 or considerably more complicated loss functions, e.g., for ranking and retrieval [21].”
Group recommendations with rank aggregation and collaborative filtering (2010, background): “prediction approaches aimed at producing better rankings [19] rather more accurate rating predictions.”
Collaborative prediction and ranking with non-random missing data (2009, background): “Unfortunately, the theory does not easily extend to the analysis of methods for learning to rank [14], or to the analysis of the ranking task itself.”