Dropout as a Low-Rank Regularizer for Matrix Factorization.
Jacopo CavazzaPietro MorerioBenjamin D. HaeffeleConnor LaneVittorio MurinoRené VidalPublished in: AISTATS (2018)
Keyphrases
- low rank
- matrix factorization
- collaborative filtering
- missing data
- low rank matrix
- semi supervised
- recommender systems
- factorization methods
- total variation
- matrix completion
- negative matrix factorization
- nonnegative matrix factorization
- trace norm
- data matrix
- multi task learning
- singular values
- sparsity constraints
- low rank matrices
- tensor factorization
- stochastic gradient descent
- data points
- active learning
- factorization method
- least squares
- nearest neighbor
- machine learning
- data sets