Files
jube-fork/Accord.Genetic/Fitness Functions/IFitnessFunction.cs
richard-churchman cc0a2bee21 Extensive modifications to the serialisation of Neural Network models to support Newtonsoft Json.net rather than Binary Serializer.
Included open-source Accord.net library code natively to the solution given the deprecated state of that project.  Accord.net will be supported inside Jube for own needs (such as upgrading to .Net 7+).

Accord.net libraries modified to compile to .Net 6.

Modified Accord.net libraries to better support Newtonsoft Json.net serialisation.

Included annotations in the Accord.net libraries to house important dependency statistics. Serialised json can be used independently of Jube to recall a Neural Network.

Removal of Binary Serializer totally as a precursor to the projects support of .Net 7+.

Extensive bug fixing in crucial functions of Neural Network training and recall.  While this part of the system remains in Alpha, it is substantially more fit for purpose.
2023-12-24 10:22:22 +01:00

38 lines
1.1 KiB
C#

// AForge Genetic Library
// AForge.NET framework
// http://www.aforgenet.com/framework/
//
// Copyright © Andrew Kirillov, 2006-2009
// [email protected]
//
namespace Accord.Genetic
{
using System;
/// <summary>
/// Fitness function interface.
/// </summary>
///
/// <remarks>The interface should be implemented by all fitness function
/// classes, which are supposed to be used for calculation of chromosomes
/// fitness values. All fitness functions should return positive (<b>greater
/// then zero</b>) value, which indicates how good is the evaluated chromosome -
/// the greater the value, the better the chromosome.
/// </remarks>
public interface IFitnessFunction
{
/// <summary>
/// Evaluates chromosome.
/// </summary>
///
/// <param name="chromosome">Chromosome to evaluate.</param>
///
/// <returns>Returns chromosome's fitness value.</returns>
///
/// <remarks>The method calculates fitness value of the specified
/// chromosome.</remarks>
///
double Evaluate( IChromosome chromosome );
}
}