A Tight Convergence Analysis for Stochastic Gradient Descent with Delayed Updates.
Yossi ArjevaniOhad ShamirNathan SrebroPublished in: ALT (2020)
Keyphrases
- convergence analysis
- stochastic gradient descent
- loss function
- step size
- matrix factorization
- least squares
- global convergence
- convergence rate
- lower bound
- online algorithms
- random forests
- worst case
- optimality conditions
- weight vector
- upper bound
- regularization parameter
- multiple kernel learning
- global optimum
- approximation methods
- optimization methods
- collaborative filtering
- recommender systems
- pairwise
- nonnegative matrix factorization
- cost function
- support vector machine