- Timestamp:
- 08/21/19 11:06:40 (5 years ago)
- Location:
- branches/3022-FastFunctionExtraction/FFX
- Files:
-
- 1 added
- 2 edited
Legend:
- Unmodified
- Added
- Removed
-
branches/3022-FastFunctionExtraction/FFX/FastFunctionExtraction.cs
r17218 r17219 1 1 using System; 2 2 using System.Threading; 3 using System.Linq; 3 4 using HeuristicLab.Common; // required for parameters collection 4 5 using HeuristicLab.Core; // required for parameters collection … … 10 11 using HeuristicLab.Random; // MersenneTwister 11 12 using HEAL.Attic; 13 using HeuristicLab.Algorithms.DataAnalysis.Glmnet; 14 using HeuristicLab.Problems.DataAnalysis; 15 using System.Collections.Generic; 16 using HeuristicLab.Encodings.SymbolicExpressionTreeEncoding; 12 17 13 namespace HeuristicLab.Algorithms.DataAnalysis.FastFunctionExtraction { 14 // each HL item needs to have a name and a description (BasicAlgorithm is an Item) 15 // The name and description of items is shown in the GUI<<<< 16 [Item(Name = "FastFunctionExtraction", Description = "An FFX algorithm.")] 18 namespace HeuristicLab.Algorithms.DataAnalysis.FastFunctionExtraction 19 { 17 20 18 // If the algorithm should be shown in the "New..." dialog it must be creatable. Entries in the new dialog are grouped to categories and ordered by priorities 19 [Creatable(Category = CreatableAttribute.Categories.Algorithms, Priority = 999)] 21 [Item(Name = "FastFunctionExtraction", Description = "An FFX algorithm.")] 22 [Creatable(Category = CreatableAttribute.Categories.Algorithms, Priority = 999)] 23 [StorableType("689280F7-E371-44A2-98A5-FCEDF22CA343")] // for persistence (storing your algorithm to a files or transfer to HeuristicLab.Hive 24 public sealed class FastFunctionExtraction : FixedDataAnalysisAlgorithm<RegressionProblem> 25 { 26 private enum Operator { Abs, Log }; 27 private static readonly double[] exponents = { 0.5, 1, 2 }; 20 28 21 [StorableType("689280F7-E371-44A2-98A5-FCEDF22CA343")] // for persistence (storing your algorithm to a files or transfer to HeuristicLab.Hive 22 public class FastFunctionExtraction : BasicAlgorithm { 23 // This algorithm only works for BinaryProblems. 24 // Overriding the ProblemType property has the effect that only BinaryProblems can be set as problem 25 // for the algorithm in the GUI 26 public override Type ProblemType { get { return typeof(BinaryProblem); } } 27 public new BinaryProblem Problem { get { return (BinaryProblem)base.Problem; } } 28 29 #region parameters 30 // If an algorithm has parameters then we usually also add properties to access these parameters. 31 // This is not strictly required but considered good shape. 32 private IFixedValueParameter<IntValue> MaxIterationsParameter { 33 get { return (IFixedValueParameter<IntValue>)Parameters["MaxIterations"]; } 34 } 35 public int MaxIterations { 36 get { return MaxIterationsParameter.Value.Value; } 37 set { MaxIterationsParameter.Value.Value = value; } 38 } 39 #endregion 29 private const string PenaltyParameterName = "Penalty"; 30 private const string ConsiderInteractionsParameterName = "Consider Interactions"; 31 private const string ConsiderDenominationParameterName = "Consider Denomination"; 32 private const string ConsiderExponentiationParameterName = "Consider Exponentiation"; 33 private const string ConsiderNonlinearFuncsParameterName = "Consider Nonlinear functions"; 34 private const string ConsiderHingeFuncsParameterName = "Consider Hinge Functions"; 40 35 41 // createable items must have a default ctor 42 public FastFunctionExtraction() { 43 // algorithm parameters are shown in the GUI 44 Parameters.Add(new FixedValueParameter<IntValue>("MaxIterations", new IntValue(10000))); 45 } 36 #region parameters 37 public IValueParameter<BoolValue> ConsiderInteractionsParameter 38 { 39 get { return (IValueParameter<BoolValue>)Parameters[ConsiderInteractionsParameterName]; } 40 } 41 #endregion 46 42 47 // Persistence uses this ctor to improve deserialization efficiency. 48 // If we would use the default ctor instead this would completely initialize the object (e.g. creating parameters) 49 // even though the data is later overwritten by the stored data. 50 [StorableConstructor] 51 public FastFunctionExtraction(StorableConstructorFlag _) : base(_) { } 43 #region properties 44 public bool ConsiderInteractions 45 { 46 get { return ConsiderInteractionsParameter.Value.Value; } 47 set { ConsiderInteractionsParameter.Value.Value = value; } 48 } 49 #endregion 52 50 53 // Each clonable item must have a cloning ctor (deep cloning, the cloner is used to handle cyclic object references) 54 public FastFunctionExtraction(FastFunctionExtraction original, Cloner cloner) : base(original, cloner) { 55 // Don't forget to call the cloning ctor of the base class 56 // This class does not have fields, therefore we don't need to actually clone anything 57 } 51 [StorableConstructor] 52 private FastFunctionExtraction(StorableConstructorFlag _) : base(_) { } 53 public FastFunctionExtraction(FastFunctionExtraction original, Cloner cloner) : base(original, cloner) 54 { 55 // Don't forget to call the cloning ctor of the base class 56 // This class does not have fields, therefore we don't need to actually clone anything 57 } 58 public FastFunctionExtraction() : base() 59 { 60 // algorithm parameters are shown in the GUI 61 Parameters.Add(new FixedValueParameter<DoubleValue>(PenaltyParameterName, "Penalty factor (alpha) for balancing between ridge (0.0) and lasso (1.0) regression", new DoubleValue(0.5))); 62 Parameters.Add(new ValueParameter<BoolValue>(ConsiderInteractionsParameterName, "True if you want to consider interactions, otherwise false.", new BoolValue(true))); 63 Parameters.Add(new ValueParameter<BoolValue>(ConsiderDenominationParameterName, "True if you want to consider denominations, otherwise false.", new BoolValue(true))); 64 Parameters.Add(new ValueParameter<BoolValue>(ConsiderExponentiationParameterName, "True if you want to consider exponentiation, otherwise false.", new BoolValue(true))); 65 Parameters.Add(new ValueParameter<BoolValue>(ConsiderNonlinearFuncsParameterName, "True if you want to consider nonlinear functions(abs, log,...), otherwise false.", new BoolValue(true))); 66 Parameters.Add(new ValueParameter<BoolValue>(ConsiderHingeFuncsParameterName, "True if you want to consider Hinge Functions, otherwise false.", new BoolValue(true))); 67 } 58 68 59 public override IDeepCloneable Clone(Cloner cloner) { 60 return new FastFunctionExtraction(this, cloner); 61 } 69 [StorableHook(HookType.AfterDeserialization)] 70 private void AfterDeserialization() { } 62 71 63 protected override void Run(CancellationToken cancellationToken) {64 int maxIters = MaxIterations;65 var problem = Problem;66 var rand = new MersenneTwister(1234);72 public override IDeepCloneable Clone(Cloner cloner) 73 { 74 return new FastFunctionExtraction(this, cloner); 75 } 67 76 68 var bestQuality = problem.Maximization ? double.MinValue : double.MaxValue; 69 70 var curItersItem = new IntValue(); 71 var bestQualityItem = new DoubleValue(bestQuality); 72 var curItersResult = new Result("Iteration", curItersItem); 73 var bestQualityResult = new Result("Best quality", bestQualityItem); 74 Results.Add(curItersResult); 75 Results.Add(bestQualityResult); 76 77 var funcs = generateBasisFunctions(); 78 79 for (int i = 0; i < maxIters; i++) { 80 curItersItem.Value = i; 81 82 // ----------------------------- 83 // IMPLEMENT YOUR ALGORITHM HERE 84 // ----------------------------- 77 public override Type ProblemType { get { return typeof(RegressionProblem); } } 78 public new RegressionProblem Problem { get { return (RegressionProblem)base.Problem; } } 85 79 86 80 87 // this is an example for random search 88 // for a more elaborate algorithm check the source code of "HeuristicLab.Algorithms.ParameterlessPopulationPyramid" 89 var cand = new BinaryVector(problem.Length, rand); 90 var quality = problem.Evaluate(cand, rand); // calling Evaluate like this is not possible for all problems... 91 if (problem.Maximization) bestQuality = Math.Max(bestQuality, quality); 92 else bestQuality = Math.Min(quality, bestQuality); 93 bestQualityItem.Value = bestQuality; 81 protected override void Run(CancellationToken cancellationToken) 82 { 83 var basisFunctions = generateBasisFunctions(Problem.ProblemData); 84 var x = Problem.ProblemData.AllowedInputsTrainingValues; 85 List<SymbolicExpressionTree> trees = new List<SymbolicExpressionTree>(); 94 86 95 // check the cancellation token to see if the used clicked "Stop"96 if (cancellationToken.IsCancellationRequested) break;97 }98 87 99 Results.Add(new Result("Execution time", new TimeSpanValue(this.ExecutionTime))); 100 } 88 foreach (var basisFunc in basisFunctions) 89 { 90 // add tree representation of basisFunc to trees 91 trees.Add(generateSymbolicExpressionTree(basisFunc)); 92 } 101 93 102 103 private object generateBasisFunctions() 94 foreach (var tree in trees) 95 { 96 // create new data through the help of the Interpreter 97 //IEnumerable<double> responses = 98 } 99 100 var coefficientVectorSet = findCoefficientValues(basisFunctions); 101 var paretoFront = nondominatedFilter(coefficientVectorSet); 102 } 103 104 private SymbolicExpressionTree generateSymbolicExpressionTree(KeyValuePair<string, double[]> basisFunc) 104 105 { 105 106 throw new NotImplementedException(); 106 107 } 107 108 108 public override bool SupportsPause { 109 get { return false; } 109 // generate all possible models 110 private static Dictionary<string, double[]> generateBasisFunctions(IRegressionProblemData problemData) 111 { 112 var basisFunctions = generateUnivariateBases(problemData); 113 return basisFunctions; 114 } 115 116 private static Dictionary<string, double[]> generateUnivariateBases(IRegressionProblemData problemData) 117 { 118 119 var dataset = problemData.Dataset; 120 var rows = problemData.TrainingIndices; 121 var B1 = new Dictionary<string, double[]>(); 122 123 foreach (var variableName in dataset.VariableNames) 124 { 125 foreach (var exp in new[] { 0.5, 1, 2 }) 126 { 127 var name = variableName + " ** " + exp; 128 var data = dataset.GetDoubleValues(variableName, rows).Select(x => Math.Pow(x, exp)).ToArray(); 129 B1.Add(name, data); 130 foreach (Operator op in Enum.GetValues(typeof(Operator))) 131 { 132 var inner_name = op.ToString() + "(" + name + ")"; 133 var inner_data = data.Select(x => executeOperator(x, op)).ToArray(); 134 B1.Add(inner_name, inner_data); 135 } 136 } 137 } 138 139 return B1; 140 } 141 142 private static double executeOperator(double x, Operator op) 143 { 144 switch (op) 145 { 146 case Operator.Abs: 147 return x > 0 ? x : -x; 148 case Operator.Log: 149 return Math.Log10(x); 150 default: 151 throw new NotImplementedException(); 152 } 153 } 154 155 private static Dictionary<string, double[]> generateMultiVariateBases(Dictionary<string, double[]> B1) 156 { 157 var B2 = new Dictionary<string, double[]>(); 158 for(int i = 1; i <= B1.Count(); i++ ) 159 { 160 var b_i = B1.ElementAt(i); 161 for (int j = 1; j < i; i++) 162 { 163 var b_j = B1.ElementAt(j); 164 } 165 } 166 167 // return union of B1 and B2 168 return B2.Concat(B1).ToDictionary(kvp => kvp.Key, kvp => kvp.Value); 169 } 170 171 private static object findCoefficientValues(IEnumerable<KeyValuePair<string, double[]>> basisFunctions) 172 { 173 return new object(); 174 } 175 176 private static object nondominatedFilter(object coefficientVectorSet) 177 { 178 return new object(); 179 } 180 181 public override bool SupportsPause 182 { 183 get { return false; } 184 } 110 185 } 111 }112 186 } -
branches/3022-FastFunctionExtraction/FFX/FastFunctionExtraction.csproj
r17218 r17219 36 36 <HintPath>..\..\..\trunk\bin\HEAL.Attic.dll</HintPath> 37 37 </Reference> 38 <Reference Include="HeuristicLab.Algorithms.DataAnalysis-3.4, Version=3.4.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL"> 39 <SpecificVersion>False</SpecificVersion> 40 <HintPath>..\..\..\trunk\bin\HeuristicLab.Algorithms.DataAnalysis-3.4.dll</HintPath> 41 </Reference> 42 <Reference Include="HeuristicLab.Algorithms.DataAnalysis.Glmnet-3.4, Version=3.4.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL"> 43 <SpecificVersion>False</SpecificVersion> 44 <HintPath>..\..\..\trunk\bin\HeuristicLab.Algorithms.DataAnalysis.Glmnet-3.4.dll</HintPath> 45 </Reference> 38 46 <Reference Include="HeuristicLab.Collections-3.3, Version=3.3.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL"> 39 47 <SpecificVersion>False</SpecificVersion> … … 56 64 <HintPath>..\..\..\trunk\bin\HeuristicLab.Encodings.BinaryVectorEncoding-3.3.dll</HintPath> 57 65 </Reference> 66 <Reference Include="HeuristicLab.Encodings.SymbolicExpressionTreeEncoding-3.4, Version=3.4.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL"> 67 <SpecificVersion>False</SpecificVersion> 68 <HintPath>..\..\..\trunk\bin\HeuristicLab.Encodings.SymbolicExpressionTreeEncoding-3.4.dll</HintPath> 69 </Reference> 58 70 <Reference Include="HeuristicLab.Optimization-3.3, Version=3.3.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL"> 59 71 <SpecificVersion>False</SpecificVersion> … … 75 87 <HintPath>..\..\..\trunk\bin\HeuristicLab.Problems.Binary-3.3.dll</HintPath> 76 88 </Reference> 89 <Reference Include="HeuristicLab.Problems.DataAnalysis-3.4, Version=3.4.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL"> 90 <SpecificVersion>False</SpecificVersion> 91 <HintPath>..\..\..\trunk\bin\HeuristicLab.Problems.DataAnalysis-3.4.dll</HintPath> 92 </Reference> 93 <Reference Include="HeuristicLab.Problems.DataAnalysis.Symbolic.Regression-3.4, Version=3.4.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL"> 94 <SpecificVersion>False</SpecificVersion> 95 <HintPath>..\..\..\trunk\bin\HeuristicLab.Problems.DataAnalysis.Symbolic.Regression-3.4.dll</HintPath> 96 </Reference> 97 <Reference Include="HeuristicLab.Problems.Instances-3.3, Version=3.3.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec" /> 77 98 <Reference Include="HeuristicLab.Random-3.3, Version=3.3.0.0, Culture=neutral, PublicKeyToken=ba48961d6f65dcec, processorArchitecture=MSIL"> 78 99 <SpecificVersion>False</SpecificVersion> … … 89 110 </ItemGroup> 90 111 <ItemGroup> 112 <Compile Include="BasisFunction.cs" /> 91 113 <Compile Include="FastFunctionExtraction.cs" /> 114 <Compile Include="GeneralizedLinearModel.cs" /> 92 115 <Compile Include="Plugin.cs" /> 93 116 <Compile Include="Properties\AssemblyInfo.cs" />
Note: See TracChangeset
for help on using the changeset viewer.