Multilayer tensor factorization with applications to recommender systems

  title={Multilayer tensor factorization with applications to recommender systems},
  author={Xuan Bi and Annie Qu and Xiaotong Shen},
  journal={The Annals of Statistics},
Recommender systems have been widely adopted by electronic commerce and entertainment industries for individualized prediction and recommendation, which benefit consumers and improve business intelligence. In this article, we propose an innovative method, namely the recommendation engine of multilayers (REM), for tensor recommender systems. The proposed method utilizes the structure of a tensor response to integrate information from multiple modes, and creates an additional layer of nested… 

Tables from this paper

Dynamic Tensor Recommender System
This article proposes a new recommendation method, namely the dynamic tensor recommender system (DTRS), which aims particularly at forecasting future recommendation and outperforms existing methods in terms of future time forecasting.
Tensor factorization recommender systems with dependency
An innovative tensor-based recommender system, namely, the Ten- sor Factorization with Dependency (TFD), which utilizes shared factors to characterize the dependency between different modes, in addition to pairwise additive tensor factorization to integrate information among multiple modes.
Improving Sales Forecasting Accuracy: A Tensor Factorization Approach with Demand Awareness
A novel computational approach called the advanced temporal latent factor approach to sales forecasting, or ATLAS, which achieves accurate and individualized predictions for sales by building a single tensor factorization model across multiple stores and products.
Two-level monotonic multistage recommender systems
This article develops a multistage recommender system utilizing a two-level monotonic property characterizing amonotonic chain of events for personalized prediction, and derives a large-margin classifier based on a nonnegative additive latent factor model in the presence of a high percentage of missing observations.
A Personalized Recommender System with Correlation Estimation
This work proposes a new personalized recommender system method that takes into account the correlation structure of ratings by a user, and shows theoretically that incorporating the correlations structure gives higher asymptotic efficiency of the estimators compared to ignoring it.
A Low-Rank Tensor Factorization Using Implicit Similarity in Trust Relationships
A recommendation model that comprehensively considers local information and global information is proposed, combining the similarity between trust users and low-rank tensor factorization, and the conjugate gradient (CG) optimization algorithm for the model is designed.
Regularized and Smooth Double Core Tensor Factorization for Heterogeneous Data
A general tensor model suitable for data analytic tasks for heterogeneous data sets, wherein there are joint low-rank structures within groups of observations, but also discriminative structures across different groups, is introduced together with a family of smoothing loss functions.
Stochastic Low-rank Tensor Bandits for Multi-dimensional Online Decision Making
Stochastic low-rank tensor bandits are introduced, a class of bandits whose mean rewards can be represented as a low- rank tensor, and a practically Bayesian algorithm called tensor ensemble sampling for Tensor bandits with context is developed.
Dynamic Tensor Clustering
  • W. Sun, Lexin Li
  • Computer Science
    Journal of the American Statistical Association
  • 2019
A new dynamic Tensor clustering method that works for a general-order dynamic tensor, and enjoys both strong statistical guarantee and high computational efficiency is proposed.
Low-rank Tensor Bandits
This work introduces stochastic low-rank tensor bandits, a class of bandits whose mean rewards can be represented as a low- rank tensor, and proposes two learning algorithms, tensor epoch-greedy and tensor elimination, and develops finite-time regret bounds for them.


Personalized Prediction and Sparsity Pursuit in Latent Factor Models
This work factorizes a user-over-item preference matrix into a product of two matrices, each representing a user’s preference and an item preference by users, and proposes a likelihood method to seek a sparsest latent factorization, from a class of overcomplete factorizations, possibly with a high percentage of missing values.
Fast context-aware recommendations with factorization machines
This work proposes to apply Factorization Machines (FMs) to model contextual information and to provide context-aware rating predictions and shows empirically that this approach outperforms Multiverse Recommendation in prediction quality and runtime.
Collaborative Filtering beyond the User-Item Matrix
A comprehensive introduction to a large body of research, more than 200 key references, is provided, with the aim of supporting the further development of recommender systems exploiting information beyond the U-I matrix.
Gaussian process factorization machines for context-aware recommendations
This work develops a novel and powerful non-linear probabilistic algorithm using Gaussian processes which outperforms state-of-the-art context-aware recommendation methods and derives stochastic gradient descent optimization to allow scalability of the model.
Factorization Machines with libFM
Factorization approaches provide high accuracy in several important prediction problems, for example, recommender systems. However, applying factorization approaches to a new prediction problem is a
Context-Aware Recommender Systems for Learning: A Survey and Future Challenges
A context framework that identifies relevant context dimensions for TEL applications is presented and an analysis of existing TEL recommender systems along these dimensions is presented, and topics on which further research is needed are outlined.
Content‐boosted matrix factorization techniques for recommender systems
Different ways to incorporate content information directly into the matrix factorization approach of collaborative filtering not only improve recommendation accuracy, but also provide useful insights about the contents, as well as make recommendations more easily interpretable.
Statistical Significance of the Netflix Challenge
The goal is to address a statistical audience, and to provide a primarily statistical treatment of the lessons that have been learned from this remarkable set of data on collaborative filtering and recommender systems.
Content-boosted matrix factorization for recommender systems: experiments with recipe recommendation
This work describes a simple algorithm for incorporating content information directly into matrix factorization approach for collaborative filtering and presents experimental evidence using recipe data to show that this not only improves recommendation accuracy but also provides useful insights about the contents themselves that are otherwise unavailable.