2003 | OriginalPaper | Buchkapitel
Geometric Ideas in Minimum Cross-Entropy
verfasst von : L. Lore Campbell
Erschienen in: Entropy Measures, Maximum Entropy Principle and Emerging Applications
Verlag: Springer Berlin Heidelberg
Enthalten in: Professional Book Archive
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
This article reviews three geometric approaches to the understanding of the minimum cross-entropy method for estimating a probability distribution. The first approach is to regard the method as a projection based on an analogue of Pythagoras’ Theorem. The second is to regard the set of probability distributions as a differentiable manifold and to introduce a Riemannian geometry on this manifold. The third uses the idea of Hausdorff dimension to support the use of the method.