ENVITrainClassifierTask

This task trains a classifier. It updates the original classifier instead of creating a new output classifier.

The following diagrams show typical workflows where this task is used:

Example

See the following topics:

Syntax

Result = ENVITask('TrainClassifier')

Input properties (Set, Get): CLASSIFIER, EXAMPLES, TRAINER

Output properties (Get only): LOSS_PROFILE, TRAINED_CLASSIFIER

Properties marked as "Set" are those that you can set to specific values. You can also retrieve their current values any time. Properties marked as "Get" are those whose values you can retrieve but not set.

Methods

This task inherits the following methods from ENVITask:

Properties

This task inherits the following properties from ENVITask:

COMMUTE_ON_DOWNSAMPLE

CLASSIFIER (required)

Specify an input classifier object; for example, ENVISVMClassifier or ENVISoftmaxRegressionClassifier.

EXAMPLES (required)

This is a reference to the input ENVIExamples object.

LOSS_PROFILE

This is an array of values showing the loss as a function of iterations. The array consists of one initial loss value plus one value per iteration. Loss is a measure of how closely the classifier algorithm fits the examples. In general, the loss decreases with each iteration until it stops changing. A loss of zero would indicate a perfect fit, but a perfect fit is not necessarily desired. The goal of training is to predict new examples correctly. A fit that is too good might result in overfitting and in a poorer classifier.

TRAINED_CLASSIFIER (optional)

This is a reference to the trained classifier object.

TRAINER (required)

Specify an input trainer object; for example, ENVIIterativeTrainer or ENVIGradientDescentTrainer.

Version History

ENVI 5.4

Introduced

API Version

3.6