Changeset 15403 for branches/MCTS-SymbReg-2796/HeuristicLab.Algorithms.DataAnalysis/3.4/MctsSymbolicRegression/MctsSymbolicRegressionAlgorithm.cs

Timestamp:

10/04/17 22:00:52 (7 years ago)

Author:

gkronber

Message:

#2796 worked on MCTS symb reg

Location:

branches/MCTS-SymbReg-2796

Files:

• . (modified) (1 prop)
• HeuristicLab.Algorithms.DataAnalysis/3.4/MctsSymbolicRegression/MctsSymbolicRegressionAlgorithm.cs (modified) (13 diffs)

branches/MCTS-SymbReg-2796/HeuristicLab.Algorithms.DataAnalysis/3.4/MctsSymbolicRegression/MctsSymbolicRegressionAlgorithm.cs

@@ -22 +22 @@
 using System;
 using System.Linq;
-using System.Runtime.CompilerServices;
 using System.Threading;
 using HeuristicLab.Algorithms.DataAnalysis.MctsSymbolicRegression.Policies;
@@ -36 +35 @@
 
 namespace HeuristicLab.Algorithms.DataAnalysis.MctsSymbolicRegression {
-  [Item("MCTS Symbolic Regression", "Monte carlo tree search for symbolic regression. Useful mainly as a base learner in gradient boosting.")]
+  // TODO: support pause (persisting/cloning the state)
+  [Item("MCTS Symbolic Regression", "Monte carlo tree search for symbolic regression.")]
   [StorableClass]
   [Creatable(CreatableAttribute.Categories.DataAnalysisRegression, Priority = 250)]
@@ -53 +53 @@
     private const string CreateSolutionParameterName = "CreateSolution";
     private const string PunishmentFactorParameterName = "PunishmentFactor";
-
-    private const string VariableProductFactorName = "product(xi)";
-    private const string ExpFactorName = "exp(c * product(xi))";
-    private const string LogFactorName = "log(c + sum(c*product(xi))";
-    private const string InvFactorName = "1 / (1 + sum(c*product(xi))";
-    private const string FactorSumsName = "sum of multiple terms";
+    private const string CollectParetoOptimalSolutionsParameterName = "CollectParetoOptimalSolutions";
+    private const string LambdaParameterName = "Lambda";
+
+    private const string VariableProductFactorName = "x * y * ...";
+    private const string ExpFactorName = "exp(c * x * y ...)";
+    private const string LogFactorName = "log(c + c1 x + c2 x + ...)";
+    private const string InvFactorName = "1 / (1 + c1 x + c2 x + ...)";
+    private const string FactorSumsName = "t1(x) + t2(x) + ... ";
     #endregion
 
@@ -94 +96 @@
     public IFixedValueParameter<BoolValue> CreateSolutionParameter {
       get { return (IFixedValueParameter<BoolValue>)Parameters[CreateSolutionParameterName]; }
+    }
+    public IFixedValueParameter<BoolValue> CollectParetoOptimalSolutionsParameter {
+      get { return (IFixedValueParameter<BoolValue>)Parameters[CollectParetoOptimalSolutionsParameterName]; }
+    }
+    public IFixedValueParameter<DoubleValue> LambdaParameter {
+      get { return (IFixedValueParameter<DoubleValue>)Parameters[LambdaParameterName]; }
     }
     #endregion
@@ -136 +144 @@
       get { return CreateSolutionParameter.Value.Value; }
       set { CreateSolutionParameter.Value.Value = value; }
+    }
+    public bool CollectParetoOptimalSolutions {
+      get { return CollectParetoOptimalSolutionsParameter.Value.Value; }
+      set { CollectParetoOptimalSolutionsParameter.Value.Value = value; }
+    }
+    public double Lambda {
+      get { return LambdaParameter.Value.Value; }
+      set { LambdaParameter.Value.Value = value; }
     }
     #endregion
@@ -177 +193 @@
       Parameters.Add(new FixedValueParameter<IntValue>(ConstantOptimizationIterationsParameterName,
         "Number of iterations for constant optimization. A small number of iterations should be sufficient for most models. " +
-        "Set to 0 to disable constants optimization.", new IntValue(10)));
+        "Set to 0 to let the algorithm stop automatically when it converges. Set to -1 to disable constants optimization.", new IntValue(10)));
       Parameters.Add(new FixedValueParameter<BoolValue>(ScaleVariablesParameterName,
-        "Set to true to scale all input variables to the range [0..1]", new BoolValue(false)));
+        "Set to true to all input variables to the range [0..1]", new BoolValue(true)));
       Parameters[ScaleVariablesParameterName].Hidden = true;
       Parameters.Add(new FixedValueParameter<DoubleValue>(PunishmentFactorParameterName, "Estimations of models can be bounded. The estimation limits are calculated in the following way (lb = mean(y) - punishmentFactor*range(y), ub = mean(y) + punishmentFactor*range(y))", new DoubleValue(10)));
@@ -187 +203 @@
       Parameters[UpdateIntervalParameterName].Hidden = true;
       Parameters.Add(new FixedValueParameter<BoolValue>(CreateSolutionParameterName,
-        "Flag that indicates if a solution should be produced at the end of the run", new BoolValue(true)));
+        "Optionally produce a solution at the end of the run", new BoolValue(true)));
       Parameters[CreateSolutionParameterName].Hidden = true;
+
+      Parameters.Add(new FixedValueParameter<BoolValue>(CollectParetoOptimalSolutionsParameterName,
+        "Optionally collect a set of Pareto-optimal solutions minimizing error and complexity.", new BoolValue(false)));
+      Parameters[CollectParetoOptimalSolutionsParameterName].Hidden = true;
+
+      Parameters.Add(new FixedValueParameter<DoubleValue>(LambdaParameterName,
+        "Lambda is the factor for the regularization term in the objective function (Obj = (y - f(x,p))² + lambda * |p|²)", new DoubleValue(0.0)));
     }
 
@@ -195 +218 @@
     }
 
+    // TODO: support pause and restart
     protected override void Run(CancellationToken cancellationToken) {
       // Set up the algorithm
       if (SetSeedRandomly) Seed = new System.Random().Next();
+      var collectPareto = CollectParetoOptimalSolutions;
 
       // Set up the results display
@@ -229 +254 @@
       var gradEvals = new IntValue();
       Results.Add(new Result("Gradient evaluations", gradEvals));
-      var paretoBestModelsResult = new Result("ParetoBestModels", typeof(ItemList<ISymbolicRegressionSolution>));
-      Results.Add(paretoBestModelsResult);
-
+
+      Result paretoBestModelsResult = new Result("ParetoBestModels", typeof(ItemList<ISymbolicRegressionSolution>));
+      if (collectPareto) {
+        Results.Add(paretoBestModelsResult);
+      }
 
       // same as in SymbolicRegressionSingleObjectiveProblem
@@ -246 +273 @@
       var problemData = (IRegressionProblemData)Problem.ProblemData.Clone();
       if (!AllowedFactors.CheckedItems.Any()) throw new ArgumentException("At least on type of factor must be allowed");
-      var state = MctsSymbolicRegressionStatic.CreateState(problemData, (uint)Seed, MaxVariableReferences, ScaleVariables, ConstantOptimizationIterations,
-        Policy,
+      var state = MctsSymbolicRegressionStatic.CreateState(problemData, (uint)Seed, MaxVariableReferences, ScaleVariables, 
+        ConstantOptimizationIterations, Lambda,
+        Policy, collectPareto,
         lowerLimit, upperLimit,
         allowProdOfVars: AllowedFactors.CheckedItems.Any(s => s.Value.Value == VariableProductFactorName),
@@ -261 +289 @@
       double curBestQ = 0.0;
       int n = 0;
+
+      // cancelled before we acutally started
+      cancellationToken.ThrowIfCancellationRequested();
+
       // Loop until iteration limit reached or canceled.
-      for (int i = 0; i < Iterations && !state.Done; i++) {
-        cancellationToken.ThrowIfCancellationRequested();
-
+      for (int i = 0; i < Iterations && !state.Done && !cancellationToken.IsCancellationRequested; i++) {
         var q = MctsSymbolicRegressionStatic.MakeStep(state);
         sumQ += q; // sum of qs in the last updateinterval iterations
@@ -286 +316 @@
           totalRollouts.Value = state.TotalRollouts;
 
-          paretoBestModelsResult.Value = new ItemList<ISymbolicRegressionSolution>(state.ParetoBestModels);
+          if (collectPareto) {
+            paretoBestModelsResult.Value = new ItemList<ISymbolicRegressionSolution>(state.ParetoBestModels);
+          }
 
           table.Rows["Best quality"].Values.Add(bestQuality.Value);
@@ -296 +328 @@
       }
 
-      // final results
+      // final results (assumes that at least one iteration was calculated)
       if (n > 0) {
         if (bestQ > bestQuality.Value) {