[6577] | 1 | #region License Information
|
---|
| 2 | /* HeuristicLab
|
---|
| 3 | * Copyright (C) 2002-2011 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 |
|
---|
| 22 | using System;
|
---|
| 23 | using System.Collections.Generic;
|
---|
| 24 | using System.Linq;
|
---|
| 25 | using HeuristicLab.Common;
|
---|
| 26 | using HeuristicLab.Core;
|
---|
| 27 | using HeuristicLab.Data;
|
---|
| 28 | using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding;
|
---|
| 29 | using HeuristicLab.Optimization;
|
---|
| 30 | using HeuristicLab.Persistence.Default.CompositeSerializers.Storable;
|
---|
| 31 | using HeuristicLab.Problems.DataAnalysis;
|
---|
| 32 | using HeuristicLab.Problems.DataAnalysis.Symbolic;
|
---|
| 33 | using HeuristicLab.Problems.DataAnalysis.Symbolic.Regression;
|
---|
| 34 | using HeuristicLab.Parameters;
|
---|
| 35 |
|
---|
| 36 | namespace HeuristicLab.Algorithms.DataAnalysis {
|
---|
| 37 | /// <summary>
|
---|
| 38 | /// Neural network regression data analysis algorithm.
|
---|
| 39 | /// </summary>
|
---|
| 40 | [Item("Neural Network Regression", "Neural network regression data analysis algorithm (wrapper for ALGLIB).")]
|
---|
| 41 | [Creatable("Data Analysis")]
|
---|
| 42 | [StorableClass]
|
---|
| 43 | public sealed class NeuralNetworkRegression : FixedDataAnalysisAlgorithm<IRegressionProblem> {
|
---|
[6578] | 44 | private const string DecayParameterName = "Decay";
|
---|
| 45 | private const string HiddenLayersParameterName = "HiddenLayers";
|
---|
| 46 | private const string NodesInFirstHiddenLayerParameterName = "NodesInFirstHiddenLayer";
|
---|
| 47 | private const string NodesInSecondHiddenLayerParameterName = "NodesInSecondHiddenLayer";
|
---|
| 48 | private const string RestartsParameterName = "Restarts";
|
---|
[6577] | 49 | private const string NeuralNetworkRegressionModelResultName = "Neural network regression solution";
|
---|
[6578] | 50 |
|
---|
| 51 | #region parameter properties
|
---|
| 52 | public IFixedValueParameter<DoubleValue> DecayParameter {
|
---|
| 53 | get { return (IFixedValueParameter<DoubleValue>)Parameters[DecayParameterName]; }
|
---|
| 54 | }
|
---|
| 55 | public ConstrainedValueParameter<IntValue> HiddenLayersParameter {
|
---|
| 56 | get { return (ConstrainedValueParameter<IntValue>)Parameters[HiddenLayersParameterName]; }
|
---|
| 57 | }
|
---|
| 58 | public IFixedValueParameter<IntValue> NodesInFirstHiddenLayerParameter {
|
---|
| 59 | get { return (IFixedValueParameter<IntValue>)Parameters[NodesInFirstHiddenLayerParameterName]; }
|
---|
| 60 | }
|
---|
| 61 | public IFixedValueParameter<IntValue> NodesInSecondHiddenLayerParameter {
|
---|
| 62 | get { return (IFixedValueParameter<IntValue>)Parameters[NodesInSecondHiddenLayerParameterName]; }
|
---|
| 63 | }
|
---|
| 64 | public IFixedValueParameter<IntValue> RestartsParameter {
|
---|
| 65 | get { return (IFixedValueParameter<IntValue>)Parameters[RestartsParameterName]; }
|
---|
| 66 | }
|
---|
| 67 | #endregion
|
---|
| 68 |
|
---|
| 69 | #region properties
|
---|
| 70 | public double Decay {
|
---|
| 71 | get { return DecayParameter.Value.Value; }
|
---|
| 72 | set {
|
---|
| 73 | if (value < 0.001 || value > 100) throw new ArgumentException("The decay parameter should be set to a value between 0.001 and 100.", "Decay");
|
---|
| 74 | DecayParameter.Value.Value = value;
|
---|
| 75 | }
|
---|
| 76 | }
|
---|
| 77 | public int HiddenLayers {
|
---|
| 78 | get { return HiddenLayersParameter.Value.Value; }
|
---|
| 79 | set {
|
---|
| 80 | if (value < 0 || value > 2) throw new ArgumentException("The number of hidden layers should be set to 0, 1, or 2.", "HiddenLayers");
|
---|
| 81 | HiddenLayersParameter.Value = (from v in HiddenLayersParameter.ValidValues
|
---|
| 82 | where v.Value == value
|
---|
| 83 | select v)
|
---|
| 84 | .Single();
|
---|
| 85 | }
|
---|
| 86 | }
|
---|
| 87 | public int NodesInFirstHiddenLayer {
|
---|
| 88 | get { return NodesInFirstHiddenLayerParameter.Value.Value; }
|
---|
| 89 | set {
|
---|
| 90 | if (value < 1) throw new ArgumentException("The number of nodes in the first hidden layer must be at least one.", "NodesInFirstHiddenLayer");
|
---|
| 91 | NodesInFirstHiddenLayerParameter.Value.Value = value;
|
---|
| 92 | }
|
---|
| 93 | }
|
---|
| 94 | public int NodesInSecondHiddenLayer {
|
---|
| 95 | get { return NodesInSecondHiddenLayerParameter.Value.Value; }
|
---|
| 96 | set {
|
---|
| 97 | if (value < 1) throw new ArgumentException("The number of nodes in the first second layer must be at least one.", "NodesInSecondHiddenLayer");
|
---|
| 98 | NodesInSecondHiddenLayerParameter.Value.Value = value;
|
---|
| 99 | }
|
---|
| 100 | }
|
---|
| 101 | public int Restarts {
|
---|
| 102 | get { return RestartsParameter.Value.Value; }
|
---|
| 103 | set {
|
---|
| 104 | if (value < 0) throw new ArgumentException("The number of restarts must be positive.", "Restarts");
|
---|
| 105 | RestartsParameter.Value.Value = value;
|
---|
| 106 | }
|
---|
| 107 | }
|
---|
| 108 | #endregion
|
---|
| 109 |
|
---|
| 110 |
|
---|
[6577] | 111 | [StorableConstructor]
|
---|
| 112 | private NeuralNetworkRegression(bool deserializing) : base(deserializing) { }
|
---|
| 113 | private NeuralNetworkRegression(NeuralNetworkRegression original, Cloner cloner)
|
---|
| 114 | : base(original, cloner) {
|
---|
| 115 | }
|
---|
| 116 | public NeuralNetworkRegression()
|
---|
| 117 | : base() {
|
---|
[6578] | 118 | var validHiddenLayerValues = new ItemSet<IntValue>(new IntValue[] { new IntValue(0), new IntValue(1), new IntValue(2) });
|
---|
| 119 | var selectedHiddenLayerValue = (from v in validHiddenLayerValues
|
---|
| 120 | where v.Value == 1
|
---|
| 121 | select v)
|
---|
| 122 | .Single();
|
---|
| 123 | Parameters.Add(new FixedValueParameter<DoubleValue>(DecayParameterName, "The decay parameter for the training phase of the neural network. This parameter determines the strengh of regularization and should be set to a value between 0.001 (weak regularization) to 100 (very strong regularization). The correct value should be determined via cross-validation.", new DoubleValue(1)));
|
---|
| 124 | Parameters.Add(new ConstrainedValueParameter<IntValue>(HiddenLayersParameterName, "The number of hidden layers for the neural network (0, 1, or 2)", validHiddenLayerValues, selectedHiddenLayerValue));
|
---|
| 125 | Parameters.Add(new FixedValueParameter<IntValue>(NodesInFirstHiddenLayerParameterName, "The number of nodes in the first hidden layer. This value is not used if the number of hidden layers is zero.", new IntValue(10)));
|
---|
| 126 | Parameters.Add(new FixedValueParameter<IntValue>(NodesInSecondHiddenLayerParameterName, "The number of nodes in the second hidden layer. This value is not used if the number of hidden layers is zero or one.", new IntValue(10)));
|
---|
| 127 | Parameters.Add(new FixedValueParameter<IntValue>(RestartsParameterName, "The number of restarts for learning.", new IntValue(2)));
|
---|
| 128 |
|
---|
[6577] | 129 | Problem = new RegressionProblem();
|
---|
| 130 | }
|
---|
| 131 | [StorableHook(HookType.AfterDeserialization)]
|
---|
| 132 | private void AfterDeserialization() { }
|
---|
| 133 |
|
---|
| 134 | public override IDeepCloneable Clone(Cloner cloner) {
|
---|
| 135 | return new NeuralNetworkRegression(this, cloner);
|
---|
| 136 | }
|
---|
| 137 |
|
---|
| 138 | #region neural network
|
---|
| 139 | protected override void Run() {
|
---|
| 140 | double rmsError, avgRelError;
|
---|
[6578] | 141 | var solution = CreateNeuralNetworkRegressionSolution(Problem.ProblemData, HiddenLayers, NodesInFirstHiddenLayer, NodesInSecondHiddenLayer, Decay, Restarts, out rmsError, out avgRelError);
|
---|
[6577] | 142 | Results.Add(new Result(NeuralNetworkRegressionModelResultName, "The neural network regression solution.", solution));
|
---|
| 143 | Results.Add(new Result("Root mean square error", "The root of the mean of squared errors of the neural network regression solution on the training set.", new DoubleValue(rmsError)));
|
---|
| 144 | Results.Add(new Result("Average relative error", "The average of relative errors of the neural network regression solution on the training set.", new PercentValue(avgRelError)));
|
---|
| 145 | }
|
---|
| 146 |
|
---|
| 147 | public static IRegressionSolution CreateNeuralNetworkRegressionSolution(IRegressionProblemData problemData, int nLayers, int nHiddenNodes1, int nHiddenNodes2, double decay, int restarts,
|
---|
| 148 | out double rmsError, out double avgRelError) {
|
---|
| 149 | Dataset dataset = problemData.Dataset;
|
---|
| 150 | string targetVariable = problemData.TargetVariable;
|
---|
| 151 | IEnumerable<string> allowedInputVariables = problemData.AllowedInputVariables;
|
---|
| 152 | IEnumerable<int> rows = problemData.TrainingIndizes;
|
---|
| 153 | double[,] inputMatrix = AlglibUtil.PrepareInputMatrix(dataset, allowedInputVariables.Concat(new string[] { targetVariable }), rows);
|
---|
| 154 | if (inputMatrix.Cast<double>().Any(x => double.IsNaN(x) || double.IsInfinity(x)))
|
---|
| 155 | throw new NotSupportedException("Neural network regression does not support NaN or infinity values in the input dataset.");
|
---|
| 156 |
|
---|
| 157 | double targetMin = problemData.Dataset.GetEnumeratedVariableValues(targetVariable).Min();
|
---|
| 158 | targetMin = targetMin - targetMin * 0.1; // -10%
|
---|
| 159 | double targetMax = problemData.Dataset.GetEnumeratedVariableValues(targetVariable).Max();
|
---|
| 160 | targetMax = targetMax + targetMax * 0.1; // + 10%
|
---|
| 161 |
|
---|
| 162 | alglib.multilayerperceptron multiLayerPerceptron = null;
|
---|
| 163 | if (nLayers == 0) {
|
---|
| 164 | alglib.mlpcreater0(allowedInputVariables.Count(), 1, targetMin, targetMax, out multiLayerPerceptron);
|
---|
| 165 | } else if (nLayers == 1) {
|
---|
| 166 | alglib.mlpcreater1(allowedInputVariables.Count(), nHiddenNodes1, 1, targetMin, targetMax, out multiLayerPerceptron);
|
---|
| 167 | } else if (nLayers == 2) {
|
---|
| 168 | alglib.mlpcreater2(allowedInputVariables.Count(), nHiddenNodes1, nHiddenNodes2, 1, targetMin, targetMax, out multiLayerPerceptron);
|
---|
| 169 | } else throw new ArgumentException("Number of layers must be zero, one, or two.", "nLayers");
|
---|
| 170 | alglib.mlpreport rep;
|
---|
| 171 | int nRows = inputMatrix.GetLength(0);
|
---|
| 172 |
|
---|
| 173 | int info;
|
---|
| 174 | // using mlptrainlm instead of mlptraines or mlptrainbfgs because only one parameter is necessary
|
---|
| 175 | alglib.mlptrainlm(multiLayerPerceptron, inputMatrix, nRows, decay, restarts, out info, out rep);
|
---|
| 176 | if (info != 2) throw new ArgumentException("Error in calculation of neural network regression solution");
|
---|
| 177 |
|
---|
| 178 | rmsError = alglib.mlprmserror(multiLayerPerceptron, inputMatrix, nRows);
|
---|
[6579] | 179 | avgRelError = alglib.mlpavgrelerror(multiLayerPerceptron, inputMatrix, nRows);
|
---|
[6577] | 180 |
|
---|
| 181 | return new NeuralNetworkRegressionSolution(problemData, new NeuralNetworkModel(multiLayerPerceptron, targetVariable, allowedInputVariables));
|
---|
| 182 | }
|
---|
| 183 | #endregion
|
---|
| 184 | }
|
---|
| 185 | }
|
---|