From Biased Selective Labels to Pseudo-Labels: An Expectation-Maximization Framework for Learning from Biased Decisions.
Trenton ChangJenna WiensPublished in: CoRR (2024)
Keyphrases
- minimax probability machine
- learning process
- learning algorithm
- learning mechanism
- main contribution
- sample selection bias
- distributed learning
- learning scheme
- learning systems
- em algorithm
- expectation maximization
- online learning
- pairwise
- high quality
- decision making
- maximum likelihood
- decision makers
- class labels
- multi label
- learning tasks
- design principles
- multi task
- probabilistic model
- positive examples
- reinforcement learning