Quantum mechanics

Algorithm

Gaussian

Pure mathematics

Factorization

Materials science

Theoretical computer science

Gradient descent

Probabilistic logic

Sparse matrix

MovieLens

Eigenvalues and eigenvectors

Scalability

Computer science

Factor (programming language)

Matrix (chemical analysis)

Physics

Collaborative filtering

Database

Mathematics

Artificial intelligence

Composite material

Programming language

Subspace topology

Recommender system

Artificial neural network

Matrix decomposition

Algebra over a field

Machine learning

Knowledge-Based Systems

Matrix factorization (MF) methods have proven as efficient and scalable approaches for collaborative filtering problems. Numerous existing MF methods rely heavily on explicit feedback. Typically, these data types may be extremely sparse; therefore, these methods can perform poorly. In order to address these challenges, we propose a latent factor model based on probabilistic MF, by incorporating implicit feedback as complementary information. Specifically, the explicit and implicit feedback matrices are decomposed into a shared subspace simultaneously. Then, the latent factor vectors are jointly optimized using a gradient descent algorithm. The experimental results using the MovieLens datasets demonstrate that the proposed algorithm outperforms the baselines.

Matrix factorization for recommendation with explicit and implicit feedback