Improved Update Rule and Sampling of Stochastic Gradient Descent with Extreme Early Stopping for Support Vector Machines.
Marcin OrchelJohan A. K. SuykensPublished in: LOD (2021)
Keyphrases
- stochastic gradient descent
- early stopping
- loss function
- support vector
- support vector machine
- least squares
- matrix factorization
- importance sampling
- step size
- random forests
- logistic regression
- svm classifier
- regularization parameter
- weight vector
- linear svm
- hyperplane
- random sampling
- cross validation
- online algorithms
- monte carlo
- pairwise
- feature selection
- kernel function
- hyperparameters
- learning models
- convergence speed
- kernel methods
- boosting algorithms
- recommender systems
- model selection
- particle filter
- markov chain