Neural Networks on C#

// AForge Neural Net Library
//
// Copyright � Andrew Kirillov, 2005-2006
// andrew.kirillov@gmail.com
//

namespace AForge.Neuro
{
	using System;

	/// <summary>
	/// Threshold activation function
	/// </summary>
	///
	/// <remarks>The class represents threshold activation function with
	/// the next expression:<br />
	/// <code>
	/// f(x) = 1, if x >= 0, otherwise 0
	/// </code>
	/// Output range of the function: <b>[0, 1]</b><br /><br />
	/// Functions graph:<br />
	/// <img src="threshold.bmp" width="242" height="172" />
	/// </remarks>
	public class ThresholdFunction : IActivationFunction
	{
		/// <summary>
		/// Initializes a new instance of the <see cref="ThresholdFunction"/> class
		/// </summary>
		public ThresholdFunction( ) { }

		/// <summary>
		/// Calculates function value
		/// </summary>
		///
		/// <param name="x">Function input value</param>
		/// 
		/// <returns>Function output value, <i>f(x)</i></returns>
		///
		/// <remarks>The method calculates function value at point <b>x</b>.</remarks>
		///
		public double Function( double x )
		{
			return ( x >= 0 ) ? 1 : 0;
		}

		/// <summary>
		/// Not supported
		/// </summary>
		/// 
		/// <param name="x">Input value</param>
		/// 
		/// <returns>Always returns 0</returns>
		/// 
		/// <remarks>The method is not supported, because it is not possible to
		/// calculate derivative of the function.</remarks>
		///
		public double Derivative( double x )
		{
			double y = Function( x );

			return 0;
		}

		/// <summary>
		/// Not supported
		/// </summary>
		/// 
		/// <param name="y">Input value</param>
		/// 
		/// <returns>Always returns 0</returns>
		/// 
		/// <remarks>The method is not supported, because it is not possible to
		/// calculate derivative of the function.</remarks>
		/// 
		public double Derivative2( double y )
		{
			return 0;
		}
	}
}