Alternating optimization method based on nonnegative matrix factorizations for deep neural networks.
Tetsuya SakuraiAkira ImakuraYuto InoueYasunori FutamuraPublished in: CoRR (2016)
Keyphrases
- optimization method
- matrix factorization
- nonnegative matrix factorization
- neural network
- genetic algorithm
- factor matrices
- tensor factorization
- data matrix
- optimization algorithm
- optimization methods
- collaborative filtering
- negative matrix factorization
- optimization process
- neural network training
- differential evolution
- simulated annealing
- global optimum
- recommender systems
- missing data
- evolutionary algorithm
- particle swarm
- artificial neural networks
- low rank
- optimization procedure
- nelder mead simplex
- least squares
- metaheuristic
- neural network model
- factorization methods
- nonlinear optimization
- back propagation
- quasi newton
- matrix multiplication
- multi objective
- objective function
- fitness function
- search space
- particle swarm optimization pso
- binary matrix
- data representation