MetAdapt: Meta-learned task-adaptive architecture for few-shot classification.
Sivan DovehEli SchwartzChao XueRogério FerisAlexander M. BronsteinRaja GiryesLeonid KarlinskyPublished in: Pattern Recognit. Lett. (2021)
Keyphrases
- pattern recognition
- classification accuracy
- automatic classification
- pattern classification
- machine learning
- learning phase
- classification algorithm
- support vector
- feature space
- feature vectors
- classification models
- image classification
- classification scheme
- decision trees
- real time
- training samples
- model selection
- feature selection
- svm classifier
- machine learning algorithms
- classification process
- classification systems
- supervised learning
- feature extraction
- learning capabilities
- cost sensitive
- unsupervised learning
- classification method
- decision rules
- benchmark datasets
- video data
- text classification
- dimensionality reduction
- training data
- image sequences
- data mining