private GaussianProcessModel(GaussianProcessModel original, Cloner cloner)
: base(original, cloner)
{
this.meanFunction = cloner.Clone(original.meanFunction);
this.covarianceFunction = cloner.Clone(original.covarianceFunction);
if (original.inputScaling != null)
{
this.inputScaling = cloner.Clone(original.inputScaling);
}
this.trainingDataset = cloner.Clone(original.trainingDataset);
this.negativeLogLikelihood = original.negativeLogLikelihood;
this.targetVariable = original.targetVariable;
this.sqrSigmaNoise = original.sqrSigmaNoise;
if (original.meanParameter != null)
{
this.meanParameter = (double[])original.meanParameter.Clone();
}
if (original.covarianceParameter != null)
{
this.covarianceParameter = (double[])original.covarianceParameter.Clone();
}
// shallow copies of arrays because they cannot be modified
this.trainingRows = original.trainingRows;
this.allowedInputVariables = original.allowedInputVariables;
this.alpha = original.alpha;
this.l = original.l;
this.x = original.x;
}
private void ObjectiveFunction(double[] x, ref double func, double[] grad, object obj)
{
// we want to optimize the model likelihood by changing the hyperparameters and also return the gradient for each hyperparameter
var data = (Tuple <IDataset, string, string[], int[], IMeanFunction, ICovarianceFunction, double[]>)obj;
var ds = data.Item1;
var targetVariable = data.Item2;
var allowedInputVariables = data.Item3;
var trainingRows = data.Item4;
var meanFunction = data.Item5;
var covarianceFunction = data.Item6;
var bestObjValue = data.Item7;
var hyperParameters = x; // the decision variable vector
try {
var model = new GaussianProcessModel(ds, targetVariable, allowedInputVariables, trainingRows, hyperParameters, meanFunction, covarianceFunction);
func = model.NegativeLogLikelihood; // mincgoptimize, so we return negative likelihood
bestObjValue[0] = Math.Max(bestObjValue[0], -func); // problem itself is a maximization problem
var gradients = model.HyperparameterGradients;
Array.Copy(gradients, grad, gradients.Length);
}
catch (ArgumentException) {
// building the GaussianProcessModel might fail, in this case we return the worst possible objective value
func = 1.0E+300;
Array.Clear(grad, 0, grad.Length);
}
}
private IItem CreateSolution()
{
var problemData = ProblemData;
var ds = problemData.Dataset;
var targetVariable = problemData.TargetVariable;
var allowedInputVariables = problemData.AllowedInputVariables.ToArray();
var trainingRows = problemData.TrainingIndices.ToArray();
lock (problemStateLocker) {
var model = new GaussianProcessModel(ds, targetVariable, allowedInputVariables, trainingRows, bestHyperParameters, (IMeanFunction)meanFunc.Clone(), (ICovarianceFunction)covFunc.Clone());
model.FixParameters();
return(model.CreateRegressionSolution((IRegressionProblemData)ProblemData.Clone()));
}
}
private GaussianProcessModel(GaussianProcessModel original, Cloner cloner)
: base(original, cloner) {
this.meanFunction = cloner.Clone(original.meanFunction);
this.covarianceFunction = cloner.Clone(original.covarianceFunction);
if (original.inputScaling != null)
this.inputScaling = cloner.Clone(original.inputScaling);
this.trainingDataset = cloner.Clone(original.trainingDataset);
this.negativeLogLikelihood = original.negativeLogLikelihood;
this.sqrSigmaNoise = original.sqrSigmaNoise;
if (original.meanParameter != null) {
this.meanParameter = (double[])original.meanParameter.Clone();
}
if (original.covarianceParameter != null) {
this.covarianceParameter = (double[])original.covarianceParameter.Clone();
}
// shallow copies of arrays because they cannot be modified
this.trainingRows = original.trainingRows;
this.allowedInputVariables = original.allowedInputVariables;
this.alpha = original.alpha;
this.l = original.l;
this.x = original.x;
}
