Discriminative variable selection for clustering with the sparse Fisher-EM algorithm.
Charles BouveyronCamille Brunet-SaumardPublished in: Comput. Stat. (2014)
Keyphrases
- em algorithm
- variable selection
- high dimensional
- expectation maximization
- group lasso
- model based clustering
- unsupervised learning
- mixture modeling
- mixture model
- high dimensional data
- maximum likelihood
- k means
- parameter estimation
- cross validation
- model selection
- density estimation
- gaussian mixture model
- maximum likelihood estimation
- sparsity inducing
- generative model
- clustering algorithm
- high dimensionality
- linear models
- dimension reduction
- feature selection
- clustering method
- structured sparsity
- data points
- feature extraction
- low dimensional
- maximum a posteriori
- hyperparameters
- document clustering
- hierarchical clustering
- nearest neighbor
- probabilistic model
- selection criterion
- cluster analysis
- least squares
- object recognition
- dimensionality reduction
- sparse representation