Powered Dirichlet Process for Controlling the Importance of "Rich-Get-Richer" Prior Assumptions in Bayesian Clustering.
Gaël Poux-MédardJulien VelcinSabine LoudcherPublished in: CoRR (2021)
Keyphrases
- dirichlet process
- nonparametric bayesian
- variational bayesian
- dirichlet process mixture models
- gaussian processes
- bayesian model
- posterior distribution
- multi task learning
- mixture model
- mixture modeling
- markov chain monte carlo
- chinese restaurant process
- clustering algorithm
- sampling algorithm
- variational inference
- bayesian framework
- gaussian process
- bayesian inference
- prior distribution
- unsupervised learning
- probability distribution
- variational methods
- bayesian methods
- k means
- latent variables
- prior knowledge
- latent dirichlet allocation
- density estimation
- posterior probability
- generative model
- maximum likelihood
- learning tasks
- kernel methods
- maximum a posteriori
- high order
- parameter estimation
- theoretical analysis
- text mining