High finite-sample efficiency and robustness based on distance-constrained maximum likelihood.
Ricardo A. MaronnaVictor J. YohaiPublished in: Comput. Stat. Data Anal. (2015)
Keyphrases
- maximum likelihood
- finite sample
- sample size
- statistical learning theory
- error bounds
- nearest neighbor
- uniform convergence
- parzen window
- expectation maximization
- distance measure
- maximum a posteriori
- euclidean distance
- mixture model
- em algorithm
- gaussian mixture model
- learning theory
- hyperparameters
- generalization bounds
- distance function