Semantic and Generalized Entropy Loss Functions for Semi-Supervised Deep Learning.
Krzysztof GajowniczekYitao LiangTal FriedmanTomasz ZabkowskiGuy Van den BroeckPublished in: Entropy (2020)
Keyphrases
- loss function
- deep learning
- semi supervised
- unsupervised learning
- pairwise
- pairwise constraints
- partially labeled data
- weakly supervised
- machine learning
- semantic information
- support vector
- convex loss functions
- reproducing kernel hilbert space
- semi supervised learning
- domain specific
- labeled data
- mental models
- active learning
- semantic similarity
- supervised learning
- multi class
- natural language
- maximum likelihood
- semantic features
- dimensionality reduction
- principal component analysis