Enhance prototypical networks with hybrid attention and confusing loss function for few-shot relation classification.
Yibing LiZuchang MaLisheng GaoYichen WuFei XieXiaoye RenPublished in: Neurocomputing (2022)
Keyphrases
- loss function
- support vector
- hinge loss
- pairwise
- learning to rank
- bayes rule
- image classification
- decision trees
- stochastic gradient descent
- support vector machine
- pattern classification
- benchmark datasets
- classification accuracy
- logistic regression
- class labels
- boosting algorithms
- risk minimization
- classification algorithm
- visual features
- model selection
- feature set
- empirical risk
- feature selection
- supervised learning
- feature extraction
- regularization term
- active learning
- svm classifier
- training samples