Preconditioning Matters: Fast Global Convergence of Non-convex Matrix Factorization via Scaled Gradient Descent.
Xixi JiaHailin WangJiangjun PengXiangchu FengDeyu MengPublished in: NeurIPS (2023)
Keyphrases
- matrix factorization
- global convergence
- conjugate gradient
- convergence rate
- globally convergent
- stochastic gradient descent
- convergence speed
- collaborative filtering
- convergence analysis
- global optimum
- low rank
- optimization methods
- recommender systems
- convex minimization
- primal dual
- missing data
- convex optimization
- objective function
- factorization methods
- step size
- negative matrix factorization
- nonnegative matrix factorization
- loss function
- tensor factorization
- low rank matrix
- search space
- training algorithm
- cost function
- particle swarm
- differential evolution
- optimization method
- newton method