The Devil is in the Details: On Models and Training Regimes for Few-Shot Intent Classification.
Mohsen MesgarThy Thy TranGoran GlavasIryna GurevychPublished in: CoRR (2022)
Keyphrases
- pattern recognition
- classification models
- training set
- machine learning
- machine learning methods
- supervised learning
- classification accuracy
- model selection
- training process
- support vector
- training samples
- decision trees
- training phase
- automatic classification
- benchmark datasets
- image classification
- statistical models
- classification scheme
- probabilistic model
- text classifiers
- video sequences
- structured prediction
- feature selection
- learning algorithm
- discriminative classifiers
- pattern classification
- video content
- avoid overfitting
- classification algorithm
- class labels
- machine learning algorithms
- video data
- support vector machine svm
- text categorization
- support vector machine
- feature vectors
- image retrieval
- training data
- feature extraction
- multimedia