Bi-level Optimization for hyperparameters in Nonnegative Matrix Factorizations.
Nicoletta Del BuonoFlavia EspositoLaura SelicatoRafal ZdunekPublished in: CoRR (2022)
Keyphrases
- bi level
- hyperparameters
- matrix factorization
- nonnegative matrix factorization
- model selection
- gray scale
- cross validation
- bayesian inference
- tensor factorization
- noise level
- em algorithm
- closed form
- prior information
- stochastic gradient descent
- factor matrices
- negative matrix factorization
- regularization parameter
- gaussian process
- bayesian framework
- random sampling
- support vector
- incomplete data
- maximum likelihood
- grid search
- maximum a posteriori
- gaussian processes
- collaborative filtering
- parameter space
- sample size
- probabilistic model
- incremental learning
- noisy images
- recommender systems
- objective function