Perturbative construction of mean-field equations in extensive-rank matrix factorization and denoising.
Antoine MaillardFlorent KrzakalaMarc MézardLenka ZdeborováPublished in: CoRR (2021)
Keyphrases
- matrix factorization
- denoising
- collaborative filtering
- missing data
- low rank
- negative matrix factorization
- recommender systems
- image denoising
- factor analysis
- nonnegative matrix factorization
- factorization methods
- total variation
- data sparsity
- natural images
- data matrix
- tensor factorization
- probabilistic matrix factorization
- markov networks
- markov random field
- implicit feedback
- stochastic gradient descent
- latent factors
- image processing
- closed form
- em algorithm
- multiscale
- bayesian networks
- item recommendation