Maximizing the Bregman divergence from a Bregman family.
Johannes RauhFrantisek MatúsPublished in: CoRR (2020)
Keyphrases
- bregman divergences
- cost sensitive
- special case
- information theoretic
- maximum entropy
- learning theory
- relative entropy
- mahalanobis distance
- loss function
- exponential family
- nearest neighbor
- kl divergence
- theoretical guarantees
- boosting algorithms
- nonnegative matrix factorization
- squared euclidean distance
- negative matrix factorization
- mutual information
- data dependent
- binary classification
- noise model
- density estimation
- closed form
- worst case
- knn