A Theoretical Characterization of Semi-supervised Learning with Self-training for Gaussian Mixture Models.
Samet OymakTalha Cihad GulcuPublished in: AISTATS (2021)
Keyphrases
- semi supervised learning
- gaussian mixture model
- co training
- mixture model
- semi supervised
- unlabeled data
- supervised learning
- unsupervised learning
- labeled data
- semi supervised classification
- machine learning
- feature vectors
- expectation maximization
- training data
- active learning
- abundant unlabeled data
- em algorithm
- maximum likelihood
- label propagation
- density estimation
- feature space
- semi supervised learning algorithms
- graph based semi supervised learning
- manifold regularization
- generative model
- neural network
- data sets
- data points
- high dimensional
- cost sensitive
- labeled and unlabeled data