Maximum Likelihood with Bias-Corrected Calibration is Hard-To-Beat at Label Shift Adaptation.
Amr AlexandariAnshul KundajeAvanti ShrikumarPublished in: ICML (2020)
Keyphrases
- maximum likelihood
- speaker adaptation
- camera calibration
- em algorithm
- parameter estimation
- expectation maximization
- parameter estimates
- multi label
- maximum a posteriori
- automatic speech recognition
- image labeling
- speech recognition
- radially symmetric
- d objects
- small animal
- adaptation strategies
- gaze estimation
- calibration method
- synthetic aperture radar
- bundle adjustment
- gaussian distribution