Dirichlet Process — how does it help clustering?

hard

Answer

$Bayesian nonparametric: number of clusters is inferred rather than pre-specified.$
$New points can start new clusters with a probability depending on concentration parameter α (Chinese Restaurant Process metaphor: 'rich get richer' + occasional new table).$
$Fit via Gibbs / variational inference.$
$Automatic model complexity selection.$
$Foundation of DPGMM (Dirichlet Process GMM) in scikit-learn.$

Check yourself — multiple choice

Same as GMM
Bayesian nonparametric — infers number of clusters via CRP with concentration α ('rich get richer' + new tables); DPGMM in scikit-learn
Random
Not real

Dirichlet Process: infers k; CRP metaphor; foundation of DPGMM.

#clustering#density

214 interview questions in this topic.