Anzeige
Erschienen in:
2019 | OriginalPaper | Buchkapitel
3. \(\beta \)-Maxent
verfasst von : Osamu Komori, Shinto Eguchi
Erschienen in: Statistical Methods for Imbalanced Data in Ecological and Biological Studies
Verlag: Springer Japan
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
Abstract
Maxent is very popular for estimating species distributions using environmental variables such as temperature, precipitation, elevation, and soil category, all of which are closely related to the habitat of the species of interest. It is designed for estimating a probability distribution that has maximum entropy subject to the condition that the sample means of environmental variables are equal to the population means. Maxent can deal with presence-only data, for which the records of positions of the species are available but those of absence of the species are not available. Hence, this kind of data can be regarded as the extreme case of imbalance data, where observations belonging to one class (\(y=0\) or \(y=1\)) are totally missing. We investigate the Maxent from the viewpoint of divergence and extend it by introducing \(\beta \)-divergence, a variant of the more general class of U-divergence.