Learning Intact Features by Erasing-Inpainting for Few-shot Classification.
Junjie LiZilei WangXiaoming HuPublished in: AAAI (2021)
Keyphrases
- prior knowledge
- feature set
- supervised learning
- classification accuracy
- feature construction
- feature extraction
- feature space
- feature vectors
- learning algorithm
- learning phase
- classification process
- classification models
- incremental learning
- benchmark datasets
- learning process
- reinforcement learning
- low level
- unsupervised learning
- classification method
- image features
- extracted features
- support vector
- svm classifier
- class labels
- image classification
- active learning
- histogram intersection kernel
- positive and unlabeled examples
- visual object recognition
- selecting features
- supervised learning algorithms
- machine learning
- decision trees
- pattern languages
- positive examples
- feature representation
- learning models
- pattern recognition
- learning tasks