Abstract

A Maximum Entropy Approach to Species Distribution Modeling
Steven Phillips - AT&T Labs - Research Miroslav Dudik - Princeton University Robert Schapire - Princeton University
We study the problem of modeling species geographic distributions, a criticalproblem in conservation biology. We propose the use of maximum-entropytechniques for this problem, specifically, sequential-update algorithms thatcan handle a very large number of features. We describe experiments comparingmaxent with a standard distribution-modeling tool, called GARP, on a dataset containing observationdata for North American breeding birds. We also study how well maxent performsas a function of the number of training examples and training time, analyzethe use of regularization to avoid overfitting when the number of examples issmall, and explore the interpretability of models constructed using maxent.

A Maximum Entropy Approach to Species Distribution Modeling

Steven Phillips - AT&T Labs - Research
Miroslav Dudik - Princeton University
Robert Schapire - Princeton University

We study the problem of modeling species geographic distributions, a criticalproblem in conservation biology. We propose the use of maximum-entropytechniques for this problem, specifically, sequential-update algorithms thatcan handle a very large number of features. We describe experiments comparingmaxent with a standard distribution-modeling tool, called GARP, on a dataset containing observationdata for North American breeding birds. We also study how well maxent performsas a function of the number of training examples and training time, analyzethe use of regularization to avoid overfitting when the number of examples issmall, and explore the interpretability of models constructed using maxent.