A Tight Convergence Analysis for Stochastic Gradient Descent with Delayed Updates.
Yossi ArjevaniOhad ShamirNathan SrebroPublished in: CoRR (2018)
Keyphrases
- convergence analysis
- stochastic gradient descent
- step size
- least squares
- loss function
- convergence rate
- matrix factorization
- global convergence
- worst case
- random forests
- lower bound
- online algorithms
- support vector machine
- upper bound
- multiple kernel learning
- optimality conditions
- regularization parameter
- weight vector
- approximation methods
- decision trees
- logistic regression
- global optimization
- machine learning