Improving Self-supervised Dimensionality Reduction: Exploring Hyperparameters and Pseudo-Labeling Strategies.
Artur André Almeida de Macedo OliveiraMateus EspadotoRoberto HirataNina S. T. HirataAlexandru C. TeleaPublished in: VISIGRAPP (Revised Selected Papers) (2021)
Keyphrases
- hyperparameters
- dimensionality reduction
- model selection
- cross validation
- random sampling
- bayesian inference
- closed form
- prior information
- bayesian framework
- gaussian process
- em algorithm
- sample size
- maximum likelihood
- incremental learning
- noise level
- support vector
- active learning
- high dimensional data
- gaussian processes
- pattern recognition
- feature extraction
- feature selection
- low dimensional
- parameter space
- manifold learning
- high dimensional
- maximum a posteriori
- data sets
- principal component analysis
- prior knowledge
- incomplete data
- image segmentation
- learning algorithm