Class OptimalDtwClassifier

This IDistanceClassifier implementation will consider both test and training samples and claculate the threshold to separate the original and forged signatures to approximate EER. Note that this classifier is not applicable for real world scenarios. It was developed to test the theoratical boundaries of threshold based classification

Inheritance
System.Object
PipelineBase
OptimalDtwClassifier
Inherited Members
Namespace: SigStat.Common.PipelineItems.Classifiers
Assembly: SigStat.Common.dll
Syntax
public class OptimalDtwClassifier : PipelineBase, ILoggerObject, IProgress, IPipelineIO, IDistanceClassifier, IClassifier

Constructors

| Improve this Doc View Source

OptimalDtwClassifier(Func<Double[], Double[], Double>)

Initializes a new instance of the OptimalDtwClassifier class.

Declaration
public OptimalDtwClassifier(Func<double[], double[], double> distanceFunction = null)
Parameters
Type Name Description
Func<System.Double[], System.Double[], System.Double> distanceFunction

The distance function.

Properties

| Improve this Doc View Source

DistanceFunction

The function used to calculate the distance between two data points during DTW calculation

Declaration
public Func<double[], double[], double> DistanceFunction { get; set; }
Property Value
Type Description
Func<System.Double[], System.Double[], System.Double>
| Improve this Doc View Source

Features

FeatureDescriptors to consider during classification

Declaration
public List<FeatureDescriptor> Features { get; set; }
Property Value
Type Description
List<FeatureDescriptor>
| Improve this Doc View Source

Sampler

Sampler used for selecting training and test sets during a benchmark

Declaration
public Sampler Sampler { get; set; }
Property Value
Type Description
Sampler
| Improve this Doc View Source

WarpingWindowLength

Length of the warping window to be used with DTW

Declaration
public int WarpingWindowLength { get; set; }
Property Value
Type Description
System.Int32

Methods

| Improve this Doc View Source

Test(ISignerModel, Signature)

Returns a double value in the range [0..1], representing the probability of the given signature belonging to the trained model.

Declaration
public double Test(ISignerModel signerModel, Signature signature)
Parameters
Type Name Description
ISignerModel signerModel
Signature signature

The signature to test
Returns
Type Description
System.Double
| Improve this Doc View Source

Train(List<Signature>)

Trains a model based on the signatures and returns the trained model

Declaration
public ISignerModel Train(List<Signature> signatures)
Parameters
Type Name Description
List<Signature> signatures
Returns
Type Description
ISignerModel

Implements

See Also