Mitigating Overfitting in Supervised Classification from Two Unlabeled Datasets: A Consistent Risk Correction Approach.
Nan LuTianyi ZhangGang NiuMasashi SugiyamaPublished in: AISTATS (2020)
Keyphrases
- supervised classification
- supervised learning
- unsupervised learning
- risk management
- supervised classifiers
- unsupervised clustering
- land cover classification
- risk factors
- risk assessment
- decision making
- benchmark datasets
- bayesian classifiers
- unlabeled data
- training data
- remotely sensed images
- unsupervised classification
- statistical learning theory
- image segmentation
- labeled data
- semi supervised learning
- database
- classification procedure
- decision trees
- dimensionality reduction
- prior knowledge
- training set
- machine learning
- model selection
- neural network
- land cover
- semi supervised
- active learning
- data sets
- object recognition
- reinforcement learning