Mixture models for density estimation — beyond GMM.
hardAnswer
- GMM assumes Gaussian components.
- Alternatives: mixture of t-distributions (heavier tails, robust), mixture of Dirichlet (categorical), mixture density networks (NN outputs mixture parameters — good for heteroscedastic regression), Bayesian nonparametric DP-mixtures (infer k).
- Choose component distribution based on data support (bounded → beta; positive → gamma; heavy tails → t).
Check yourself — multiple choice
- Only GMM
- Mixture of t / gamma / Dirichlet / beta based on data support; MDN for heteroscedastic regression; DP-mixture for infinite components
- Random
- Same as k-means
Mixture models: GMM / t / gamma / MDN / DP-mixture — pick per data support.
#density#clustering
Practise Unsupervised Learning
214 interview questions in this topic.