SVMs, Gaussian mixtures, and their generative/discriminative fusion.
Thomas DeselaersGeorg HeigoldHermann NeyPublished in: ICPR (2008)
Keyphrases
- gaussian mixture
- discriminative classifiers
- discriminative learning
- support vector
- discriminative methods
- generative model
- em algorithm
- maximum margin
- discriminative training
- expectation maximization
- feature selection
- gaussian mixture model
- probability density function
- bayesian network classifiers
- minimum classification error
- closed form
- mixture model
- unsupervised learning
- hidden markov models
- max margin
- covariance matrix
- density estimation
- multi class
- support vector machine
- knn
- machine learning
- training set
- image processing
- latent variables
- feature space
- object categories
- information retrieval
- computer vision
- training data
- structured prediction
- lower bound
- maximum likelihood
- prior knowledge
- image patches
- probability distribution
- semi supervised