Geometry and clustering with metrics derived from separable Bregman divergences.
Erika Gomes-GonçalvesHenryk GzylFrank NielsenPublished in: CoRR (2018)
Keyphrases
- bregman divergences
- information geometry
- information theoretic
- cost sensitive
- squared euclidean distance
- maximum entropy
- clustering algorithm
- fisher information
- clustering method
- k means
- data clustering
- loss function
- kl divergence
- information theory
- mahalanobis distance
- nonnegative matrix factorization
- unsupervised learning
- exponential family
- learning theory
- nearest neighbor
- theoretical guarantees
- mutual information
- probability distribution
- high dimensional data
- maximum likelihood
- bayesian networks