/// <summary>
/// Initializes a new instance of the <see cref="AdaBoost<TModel>"/> class.
/// </summary>
///
/// <param name="model">The model to be learned.</param>
/// <param name="creationFunction">The model fitting function.</param>
///
public AdaBoost(Boost <TModel> model, ModelConstructor <TModel> creationFunction)
{
this.classifier = model;
this.Creation = creationFunction;
this.convergence = new RelativeConvergence();
}
/// <summary>
/// Initializes a new instance of the <see cref="AdaBoost<TModel>"/> class.
/// </summary>
///
/// <param name="model">The model to be learned.</param>
///
public AdaBoost(Boost <TModel> model)
{
this.classifier = model;
this.convergence = new RelativeConvergence();
}
/// <summary>
/// Initializes a new instance of the <see cref="AdaBoost<TModel>"/> class.
/// </summary>
///
/// <param name="model">The model to be learned.</param>
///
public AdaBoost(Boost <TModel> model)
{
this.Model = model;
}
/// <summary>
/// Learns a model that can map the given inputs to the given outputs.
/// </summary>
///
/// <param name="x">The model inputs.</param>
/// <param name="y">The desired outputs associated with each <paramref name="x">inputs</paramref>.</param>
/// <param name="weights">The weight of importance for each input-output pair (if supported by the learning algorithm).</param>
///
/// <returns>A model that has learned how to produce <paramref name="y"/> given <paramref name="x"/>.</returns>
///
public Boost <TModel> Learn(double[][] x, int[] y, double[] weights = null)
{
if (weights == null)
{
weights = Vector.Create(x.Length, 1.0 / x.Length);
}
if (Model == null)
{
Model = new Boost <TModel>();
}
double error = 0;
double weightSum = 0;
var predicted = new int[y.Length];
var parameters = new AdaBoostParameters();
TModel model;
do
{
// Create and train a classifier
var learner = Learner(parameters);
model = learner.Learn(x, y, weights);
if (model == null)
{
break;
}
// Determine its current accuracy
model.Decide(x, result: predicted);
error = 0;
for (int i = 0; i < predicted.Length; i++)
{
if (predicted[i] != y[i])
{
error += weights[i];
}
}
if (error >= threshold)
{
break;
}
// AdaBoost
double w = 0.5 * System.Math.Log((1.0 - error) / error);
double sum = 0;
for (int i = 0; i < weights.Length; i++)
{
double d;
if (y[i] == predicted[i])
{
d = Math.Exp(-w);
}
else
{
d = Math.Exp(+w);
}
weights[i] *= d;
sum += weights[i];
}
// Update sample weights
for (int i = 0; i < weights.Length; i++)
{
weights[i] /= sum;
}
Model.Add(w, model);
weightSum += w;
convergence.NewValue = error;
} while (!convergence.HasConverged);
// Normalize weights for confidence calculation
for (int i = 0; i < Model.Models.Count; i++)
{
Model.Models[i].Weight /= weightSum;
}
return(Model);
}
public AdaBoost(Boost <TModel> model, ModelConstructor <TModel> creationFunction)
{
this.Model = model;
this.Creation = creationFunction;
}
