Skip to main content

Faculty & Research

Close

Low-rank Matrix Factorization with Attributes

Working Paper
The authors develop a new collaborative filtering (CF) method that combines both previously known users' preferences, i.e., standard CF, as well as product/user attributes, i.e., classical function approximation, to predict a given user's interest in a particular product. This method is a generalized low rank matrix completion problem, where the authors learn a function whose inputs are pairs of vectors - the standard low rank matrix completion problem being a special case where the inputs to the function are the row and column indices of the matrix. The authors solve this generalized matrix completion problem using tensor product kernels for which they also formally generalize standard kernel properties. Benchmark experiments on movie ratings show the advantages of their generalized matrix completion method over the standard matrix completion one with no information about movies or people, as well as over standard multi-task or single task learning methods.
Faculty

Professor of Decision Sciences and Technology Management