Fix your classifier: the marginal value of training the last weight layer.
Elad HofferItay HubaraDaniel SoudryPublished in: ICLR (Poster) (2018)
Keyphrases
- training process
- training set
- training examples
- training samples
- training data
- feature selection
- decision trees
- feature selection and classifier
- training dataset
- learning algorithm
- supervised training
- svm classifier
- classifier training
- avoid overfitting
- labeled training data
- text classifiers
- classification method
- training set size
- training algorithm
- probability distribution
- support vector
- learning stage
- discriminative classifiers
- classification process
- multi layer perceptron
- training phase
- classification scheme
- multi layer
- supervised learning
- test set
- neural network
- classification performances
- sufficient training data
- text categorization
- training procedure
- support vector machine
- classification algorithm
- face detection
- feature set