Exploiting the characteristics of matrix factorization for active learning in recommender systems.
Rasoul KarimiChristoph FreudenthalerAlexandros NanopoulosLars Schmidt-ThiemePublished in: RecSys (2012)
Keyphrases
- matrix factorization
- recommender systems
- active learning
- collaborative filtering
- low rank
- cold start problem
- negative matrix factorization
- data sparsity
- factorization methods
- implicit feedback
- user preferences
- random sampling
- missing data
- factor analysis
- personalized recommendation
- user interests
- probabilistic matrix factorization
- nonnegative matrix factorization
- tensor factorization
- item recommendation
- image classification
- transfer learning
- labeled data
- variational bayesian
- recommendation quality
- machine learning
- user feedback
- user profiles
- stochastic gradient descent
- latent factors
- text mining
- latent factor models
- supervised learning
- feature space