Unsupervised Parameter Estimation of Non Linear Scaling for Improved Classification in the Dissimilarity Space.
Mauricio Orozco-AlzateRobert P. W. DuinManuele BicegoPublished in: S+SSPR (2016)
Keyphrases
- parameter estimation
- model selection
- maximum likelihood
- least squares
- supervised learning
- statistical models
- parameter estimation algorithm
- random fields
- unsupervised learning
- machine learning
- approximate inference
- expectation maximization
- markov random field
- em algorithm
- posterior distribution
- feature selection
- parameter values
- feature space
- structure learning
- support vector
- maximum likelihood estimation
- model fitting
- feature vectors
- computer vision
- optical flow
- logic programs
- parameter estimates
- estimation problems
- parameters estimation
- probability distribution