A gradient-based algorithm competitive with variational Bayesian EM for mixture of Gaussians.
Mikael KuuselaTapani RaikoAntti HonkelaJuha KarhunenPublished in: IJCNN (2009)
Keyphrases
- expectation maximization
- variational bayesian
- k means
- learning algorithm
- detection algorithm
- em algorithm
- probabilistic model
- gaussian mixture
- maximum likelihood estimation
- mixture of gaussians
- segmentation algorithm
- density estimation
- markov random field
- unsupervised learning
- background subtraction
- incomplete data