On the Promise of the Stochastic Generalized Gauss-Newton Method for Training DNNs.
Matilde GargianiAndrea ZanelliMoritz DiehlFrank HutterPublished in: CoRR (2020)
Keyphrases
- newton method
- linear svm
- quasi newton
- linear equations
- convergence analysis
- training process
- variational inequalities
- training algorithm
- regularized least squares
- quadratic programming
- logistic regression
- optimality conditions
- hyperplane
- machine learning
- sensitivity analysis
- text categorization
- training samples
- online learning
- supervised learning
- decision trees