Unsupervised dimensionality reduction via gradient-based matrix factorization with two adaptive learning rates.
Vladimir NikulinTian-Hsiang HuangPublished in: ICML Unsupervised and Transfer Learning (2012)
Keyphrases
- matrix factorization
- learning rate
- collaborative filtering
- low rank
- recommender systems
- missing data
- convergence rate
- negative matrix factorization
- factorization methods
- nonnegative matrix factorization
- learning algorithm
- covering numbers
- uniform convergence
- convergence speed
- data matrix
- gaussian kernels
- weight vector
- stochastic gradient descent
- binary matrix