Earth Imagery Segmentation on Terrain Surface with Limited Training Labels: A Semi-supervised Approach based on Physics-Guided Graph Co-Training.
Wenchong HeArpan Man SainjuZhe JiangDa YanYang ZhouPublished in: ACM Trans. Intell. Syst. Technol. (2022)
Keyphrases
- co training
- semi supervised
- label propagation
- training examples
- semi supervised learning
- unlabeled data
- partially labeled data
- labeled instances
- supervised learning
- labeled data
- labeled training data
- semi supervised classification
- labeled data for training
- labeled examples
- three dimensional
- unlabelled data
- small set of labeled
- multi view
- training process
- unlabeled examples
- training set
- pairwise constraints
- pairwise
- active learning
- label information
- single view
- labelled data
- email classification
- unlabeled samples
- semi supervised learning algorithms
- unlabeled instances
- d objects
- class labels
- training data
- weakly supervised
- multi view learning
- semantic segmentation
- supervised methods
- high resolution
- image segmentation
- training samples
- unsupervised learning
- surface reconstruction
- small number
- elevation data
- image data
- text classification
- learning algorithm
- semi supervised learning methods
- labeled and unlabeled data
- support vector
- gaussian process
- named entities
- multiple views