openModeller

This release contains a new Maximum Entropy algorithm with two training methods (Generalized Iterative Scaling and Limited-Memory Variable Metric), new command line tools (om_points to retrieve occurrence data using any of the available drivers and om_algorithm to get information about the available algorithms), and new drivers to read occurrence data from the GBIF REST Web service, TAPIR/DarwinCore providers, or openModeller serialized XML. The GARP Best Subsets algorithm now accepts the "max threads" parameter that can be used to speed up the modeling process in multi-processor machines.

This release includes a new algorithm called
AquaMaps, which was specifically designed to model
distribution of marine organisms. Two other
algorithms were removed (minimum distance and
distance to average) since Environmental Distance
now provides the same functionality. TerraLib
drivers were updated for compatibility with
TerraLib 3.2.0. Two new classes for pre-analysis
on input layers are available: Jackknife and
ChiSquare. This release also contains improvements
in command-line tools (om_pseudo, om_create and
om_project), some changes in the API, and a few
bugfixes.

This release fixes compilation issues under Windows (MSVC compiler), includes a new command line program to generate pseudo occurrences, has minor improvements in console tools (absences are now displayed in om_viewer and om_niche), and has some code cleanup.

This release fixes MSVC compilation problems, a bug in the deserialization of OM GARP, a crash in one-class SVM when input points contained absences, a bug in deserialization of environmental distance algorithm for the Mahalanobis metric, and a bug in the pseudo-absence generation of the SVM algorithm when no absences were passed as a parameter. It implements serialization/deserialization for OM GARP Best Subsets, and includes a new algorithm "Envelope Score".

This release includes a new algorithm based on Support Vector Machines (C-SVC, nu-SVC, and one-class SVM), adds support for multiple normalization techniques (two implementations are available: ScaleNormalizer and MeanVarianceNormalizer), has a new method to cancel jobs (model creation or model projection), makes Sample serialization based on the original (unnormalized) environment values, and contains a new infrastructure for unit tests using cxxtest.