Statistics: CompactClassificationNeuralNetwork

Class Definition: `CompactClassificationNeuralNetwork`

statistics: CompactClassificationNeuralNetwork

Compact neural network classification

The CompactClassificationNeuralNetwork class implements a compact version of the neural network classifier object, which can predict responses for new data using the predict method, but does not store the training data.

A compact neural network classification model is a smaller version of the full ClassificationNeuralNetwork model that does not include the training data. It consumes less memory than the full model, but cannot perform tasks that require the training data, such as cross-validation.

Create a CompactClassificationNeuralNetwork object by using the compact method on a ClassificationNeuralNetwork object.

See also: ClassificationNeuralNetwork, fitcnet

Source Code: CompactClassificationNeuralNetwork

Properties

A positive integer value specifying the number of predictors in the training dataset used for training the neural network model. This property is read-only.

A cell array of character vectors specifying the names of the predictor variables. The names are in the order in which the appear in the training dataset. This property is read-only.

A character vector specifying the name of the response variable Y. This property is read-only.

An array of unique values of the response variable Y, which has the same data types as the data in Y. This property is read-only. ClassNames can have any of the following datatypes:

Cell array of character vectors
Character array
Logical vector
Numeric vector

Specified as a function handle for transforming the classification scores. Add or change the ScoreTransform property using dot notation as in:

obj.ScoreTransform = 'function_name'
obj.ScoreTransform = @function_handle

When specified as a character vector, it can be any of the following built-in functions. Nevertherless, the ScoreTransform property always stores their function handle equivalent.

`Value`		`Description`
`"doublelogit"`		$1 ./ (1 + e×p .^ (-2××))$
`"invlogit"`		$log (× ./ (1 -×))$
`"ismax"`		Sets the score for the class with the largest score to 1, and for all other classes to 0
`"logit"`		$1 ./ (1 + e×p .^ (-×))$
`"none"`		$×$ (no transformation)
`"identity"`		$×$ (no transformation)
`"sign"`		$-1 for× < 0, 0 for× = 0, 1 for× > 0$
`"symmetric"`		$2×× + 1$
`"symmetricismax"`		Sets the score for the class with the largest score to 1, and for all other classes to -1
`"symmetriclogit"`		$2 ./ (1 + e×p .^ (-×)) - 1$

A boolean flag indicating whether the predictor data has been standardized prior to training. When true, the predictors are centered and scaled to have zero mean and unit variance. This property is read-only.

A numeric vector containing the standard deviations of the predictors used for standardization. Empty if Standardize is false. This property is read-only.

A numeric vector containing the means of the predictors used for standardization. Empty if Standardize is false. This property is read-only.

A positive integer vector specifying the sizes of the fully connected layers in the neural network model. The i-th element of LayerSizes is the number of outputs in the i-th fully connected layer of the neural network model. LayerSizes does not include the size of the final fully connected layer. This layer always has K outputs, where K is the number of classes in Y. This property is read-only.

A character vector or cell array of character vectors specifying the activation functions used in the hidden layers of the neural network. Supported activation functions include: "linear", "sigmoid", "relu", "tanh", "softmax", "lrelu", "prelu", "elu", and "gelu". This property is read-only.

A character vector specifying the activation function of the output layer of the neural network. Supported activation functions are the same as for the Activations property. This property is read-only.

A positive scalar value defining the learning rate used by the gradient descent algorithm during training. This property is read-only.

A positive integer value defining the maximum number of epochs for training the model. This property is read-only.

A structure containing the parameters used to train the neural network classifier model, including layer weights and activations as generated by the fcnntrain function. This property is read-only.

A structure containing convergence information of the neural network classifier model with the following fields:

Accuracy - The prediction accuracy at each iteration during training
TrainingLoss - The loss value recorded at each iteration during training
Time - The cumulative time taken for all iterations in seconds

This property is read-only.

A boolean flag indicating whether to print information during training. This property is read-only.

A character vector specifying the solver algorithm used to train the neural network model. Currently only "Gradient Descent" is supported. This property is read-only.

Methods

CompactClassificationNeuralNetwork: label = predict (obj, XC)
CompactClassificationNeuralNetwork: [label, score] = predict (obj, XC)

label = predict (obj, XC) returns the vector of labels predicted for the corresponding instances in XC, using the neural network model stored in the CompactClassificationNeuralNetwork model, obj.

obj must be a CompactClassificationNeuralNetwork class object.
XC must be an $M×P$ numeric matrix with the same number of features $P$ as the corresponding predictors of the neural network model in obj.

[label, score] = predict (obj, XC) also returns score, which contains the predicted class scores or posterior probabilities for each instance of the corresponding unique classes.

The score matrix contains the classification scores for each class. For each observation in XC, the predicted class label is the one with the highest score among all classes. If the ScoreTransform property is set to a transformation function, the scores are transformed accordingly before being returned.

CompactClassificationNeuralNetwork: savemodel (obj, filename)

savemodel (obj, filename) saves each property of a CompactClassificationNeuralNetwork object into an Octave binary file, the name of which is specified in filename, along with an extra variable, which defines the type classification object these variables constitute. Use loadmodel in order to load a classification object into Octave’s workspace.

Examples

 # and compare their size

 load fisheriris
 X = meas;
 Y = species;

 Mdl = fitcnet (X, Y, 'ClassNames', unique (species))
 CMdl = crossval (Mdl)

Mdl =

  ClassificationNeuralNetwork

             ResponseName: 'Y'
               ClassNames: {'setosa' 'versicolor' 'virginica'}
           ScoreTransform: 'none'
          NumObservations: 150
            NumPredictors: 4
               LayerSizes: [10]
              Activations: 'sigmoid'
    OutputLayerActivation: 'sigmoid'
                   Solver: 'Gradient Descend'
CMdl =

  ClassificationPartitionedModel object with properties:

                   BinEdges: []
      CategoricalPredictors: []
                          X: [5.1000, 3.5000, 1.4000, 0.2000; 4.9000, 3, 1.4000, 0.2000; 4.7000, 3.2000, ...]
                          Y: [150x1 cell]
                 ClassNames: [3x1 cell]
                       Cost: []
        CrossValidatedModel: 'ClassificationNeuralNetwork'
                      KFold: 10
            ModelParameters: [1x1 struct]
            NumObservations: 150
                  Partition: [1x1 cvpartition]
             PredictorNames: [1x4 cell]
                      Prior: []
               ResponseName: "Y"
             ScoreTransform: [1x1 function_handle]
                Standardize: 0
                    Trained: [10x1 cell]

Categories &

Functions List

Clustering

Clustering

Classification Classes

Classification Classes

Clustering Classes

Clustering Classes

Regression Classes

Regression Classes

Data Manipulation

Data Manipulation

Descriptive Statistics

Descriptive Statistics

Distribution Classes

Distribution Classes

Distribution Fitting

Distribution Fitting

Distribution Functions

Distribution Functions

Distribution Statistics

Distribution Statistics

Distribution Wrappers

Distribution Wrappers

Experimental Design

Experimental Design

Machine Learning

Machine Learning

Model Fitting

Model Fitting

Hypothesis Testing

Hypothesis Testing

I/O

I/O

Plotting

Plotting

Regression

Regression

Transforms

Transforms

Class Definition: CompactClassificationNeuralNetwork

statistics: CompactClassificationNeuralNetwork

Properties

NumPredictors

PredictorNames

ResponseName

ClassNames

ScoreTransform

Standardize

Sigma

Mu

LayerSizes

Activations

OutputLayerActivation

LearningRate

IterationLimit

ModelParameters

ConvergenceInfo

DisplayInfo

Solver

Methods

predict

CompactClassificationNeuralNetwork: label = predict (obj, XC)

CompactClassificationNeuralNetwork: [label, score] = predict (obj, XC)

savemodel

CompactClassificationNeuralNetwork: savemodel (obj, filename)

Examples

Class Definition: `CompactClassificationNeuralNetwork`

`NumPredictors`

`PredictorNames`

`ResponseName`

`ClassNames`

`ScoreTransform`

`Standardize`

`Sigma`

`Mu`

`LayerSizes`

`Activations`

`OutputLayerActivation`

`LearningRate`

`IterationLimit`

`ModelParameters`

`ConvergenceInfo`

`DisplayInfo`

`Solver`

`predict`

`CompactClassificationNeuralNetwork: label = predict (obj, XC)`

`CompactClassificationNeuralNetwork: [label, score] = predict (obj, XC)`

`savemodel`

`CompactClassificationNeuralNetwork: savemodel (obj, filename)`