What's New in ENVI Machine Learning 6.2

This release includes the following new and improved features.

Highlights

Separate ENVI Machine Learning installer
New tool to publish Machine Learning models to the ENVI Repository
New and updated ENVI Toolbox tools
New and updated ENVITasks
New Machine Learning Model option added to ENVI Modeler
Compatibility With Older Machine Learning Models

ENVI Machine Learning Installer

As of this release, ENVI Machine Learning is installed as a standalone module for ENVI (it is no longer installed with ENVI Deep Learning). ENVI Machine Learning is included with ENVI, so no additional license is needed to use it.

Collaborative Model Publishing Using the ENVI Repository

The new Publish Machine Learning Model tool in the Toolbox enables you to upload (publish) Machine Learning models to the ENVI Repository. The ENVI Repository is a cross-platform server that automates the management of file dependencies across multiple machines, ensuring files can be uploaded and then downloaded and used on different machines.

New Data Manager for Machine Learning Models

When you click the Browse button to select an Input Model to publish, the new data manager for Machine Learning models appears. The models you open through the Select Machine Learning Model File dialog will be remembered, and they will appear in the list in future sessions when you browse for models to publish.

You can also write a script to publish Machine Learning models to the ENVI Repository using the PublishMachineLearningModel task.

Publish Machine Learning Model provides details on how to use this tool. For additional details on the ENVI Repository, see the "ENVI Repository" topic in ENVI Help.

New and Updated ENVI Toolbox Tools

New tools:

Publish Machine Learning Model

Updated tools:

With the new Output Model parameter, you can specify a location and filename to save your model. It has been added to the following tools:

New and Updated Tasks and Routines

New tasks:

PublishMachineLearningModel: This task uploads a trained ENVI Machine Learning model package to an ENVI Repository.

Updated tasks:

The MODEL_VERSION parameter allows you to specify a semantic version format for the trained model. The OUTPUT_MODEL parameter allows you to specify a location and filename to save your model. These parameters have been added to the following tasks:

Updated routines:

ENVIMachineLearningModel: Added the VERSION keyword.

New ENVI Modeler Machine Learning Model Option

The ENVI Modeler Select Type dialog has been updated to include an option for Machine Learning Model.

For additional information on using the ENVI Modeler, see the "ENVI Modeler" topic in ENVI Help.

Compatibility With Older Machine Learning Models

ENVI Machine Learning 6.2 includes changes that will affect the compatibility of Machine Learning models created before 6.2. Older Machine Learning models will require retraining to be compatible with 6.2.