Free cookie consent management tool by TermsFeed Policy Generator

source: branches/EfficientGlobalOptimization/HeuristicLab.Algorithms.EGO/InfillCriteria/ExpectedImprovement.cs @ 14768

Last change on this file since 14768 was 14768, checked in by bwerth, 7 years ago

#2745 fixed pausing and stopping and namespaces; Added MaximalDatasetSize and rudimentary ISurrogateAlgorithm-interface

File size: 5.6 KB
Line 
1#region License Information
2/* HeuristicLab
3 * Copyright (C) 2002-2016 Heuristic and Evolutionary Algorithms Laboratory (HEAL)
4 *
5 * This file is part of HeuristicLab.
6 *
7 * HeuristicLab is free software: you can redistribute it and/or modify
8 * it under the terms of the GNU General Public License as published by
9 * the Free Software Foundation, either version 3 of the License, or
10 * (at your option) any later version.
11 *
12 * HeuristicLab is distributed in the hope that it will be useful,
13 * but WITHOUT ANY WARRANTY; without even the implied warranty of
14 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
15 * GNU General Public License for more details.
16 *
17 * You should have received a copy of the GNU General Public License
18 * along with HeuristicLab. If not, see <http://www.gnu.org/licenses/>.
19 */
20#endregion
21
22using System;
23using System.Linq;
24using HeuristicLab.Common;
25using HeuristicLab.Core;
26using HeuristicLab.Data;
27using HeuristicLab.Encodings.RealVectorEncoding;
28using HeuristicLab.Parameters;
29using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
30using HeuristicLab.Problems.DataAnalysis;
31
32// ReSharper disable once CheckNamespace
33namespace HeuristicLab.Algorithms.EGO {
34
35  [StorableClass]
36  [Item("ExpectedImprovementMeassure", "Extension of the Expected Improvement to a weighted version by ANDRAS SÓBESTER , STEPHEN J. LEARY and ANDY J. KEANE   in \n On the Design of Optimization Strategies Based on Global Response Surface Approximation Models")]
37  public class ExpectedImprovement : InfillCriterionBase {
38
39    #region ParameterNames
40    private const string ExploitationWeightParameterName = "ExploitationWeight";
41    #endregion
42
43    #region ParameterProperties
44    public IFixedValueParameter<DoubleValue> ExploitationWeightParameter
45    {
46      get { return Parameters[ExploitationWeightParameterName] as IFixedValueParameter<DoubleValue>; }
47    }
48    #endregion
49
50    #region Properties
51    private double ExploitationWeight
52    {
53      get { return ExploitationWeightParameter.Value.Value; }
54    }
55    #endregion
56
57    #region HL-Constructors, Serialization and Cloning
58    [StorableConstructor]
59    private ExpectedImprovement(bool deserializing) : base(deserializing) { }
60    [StorableHook(HookType.AfterDeserialization)]
61    private void AfterDeserialization() {
62      RegisterEventhandlers();
63    }
64    private ExpectedImprovement(ExpectedImprovement original, Cloner cloner) : base(original, cloner) {
65      RegisterEventhandlers();
66    }
67    public ExpectedImprovement() {
68      Parameters.Add(new FixedValueParameter<DoubleValue>(ExploitationWeightParameterName, "A value between 0 and 1 indicating the focus on exploration (0) or exploitation (1)", new DoubleValue(0.5)));
69      RegisterEventhandlers();
70    }
71    public override IDeepCloneable Clone(Cloner cloner) {
72      return new ExpectedImprovement(this, cloner);
73    }
74    #endregion
75
76    public override double Evaluate(IRegressionSolution solution, RealVector vector, bool maximization) {
77      if (maximization) throw new NotImplementedException("Expected Improvement for maximization not yet implemented");
78      var model = solution.Model as IConfidenceRegressionModel;
79      if (model == null) throw new ArgumentException("can not calculate EI without confidence measure");
80      var yhat = model.GetEstimation(vector);
81      var min = solution.ProblemData.TargetVariableTrainingValues.Min();
82      var s = Math.Sqrt(model.GetVariance(vector));
83      return GetEstimatedImprovement(min, yhat, s, ExploitationWeight);
84    }
85
86    public override bool Maximization(bool expensiveProblemMaximization) {
87      return true;
88    }
89
90    #region Eventhandling
91    private void RegisterEventhandlers() {
92      DeregisterEventhandlers();
93      ExploitationWeightParameter.Value.ValueChanged += OnExploitationWeightChanged;
94    }
95    private void DeregisterEventhandlers() {
96      ExploitationWeightParameter.Value.ValueChanged -= OnExploitationWeightChanged;
97    }
98    private void OnExploitationWeightChanged(object sender, EventArgs e) {
99      ExploitationWeightParameter.Value.ValueChanged -= OnExploitationWeightChanged;
100      ExploitationWeightParameter.Value.Value = Math.Max(0, Math.Min(ExploitationWeight, 1));
101      ExploitationWeightParameter.Value.ValueChanged += OnExploitationWeightChanged;
102    }
103    #endregion
104
105    #region Helpers
106    private static double GetEstimatedImprovement(double ymin, double yhat, double s, double w) {
107      if (Math.Abs(s) < double.Epsilon) return 0;
108      var val = (ymin - yhat) / s;
109      var res = w * (ymin - yhat) * StandardNormalDistribution(val) + (1 - w) * s * StandardNormalDensity(val);
110      return double.IsInfinity(res) || double.IsNaN(res) ? 0 : res;
111    }
112
113    private static double StandardNormalDensity(double x) {
114      if (Math.Abs(x) > 10) return 0;
115      return Math.Exp(-0.5 * x * x) / Math.Sqrt(2 * Math.PI);
116    }
117
118    //taken from https://www.johndcook.com/blog/2009/01/19/stand-alone-error-function-erf/
119    private static double StandardNormalDistribution(double x) {
120      if (x > 10) return 1;
121      if (x < -10) return 0;
122      const double a1 = 0.254829592;
123      const double a2 = -0.284496736;
124      const double a3 = 1.421413741;
125      const double a4 = -1.453152027;
126      const double a5 = 1.061405429;
127      const double p = 0.3275911;
128      var sign = x < 0 ? -1 : 1;
129      x = Math.Abs(x) / Math.Sqrt(2.0);
130      var t = 1.0 / (1.0 + p * x);
131      var y = 1.0 - ((((a5 * t + a4) * t + a3) * t + a2) * t + a1) * t * Math.Exp(-x * x);
132      return 0.5 * (1.0 + sign * y);
133    }
134    #endregion
135  }
136}
Note: See TracBrowser for help on using the repository browser.