Free cookie consent management tool by TermsFeed Policy Generator

Changeset 15425 for branches


Ignore:
Timestamp:
10/20/17 12:38:06 (7 years ago)
Author:
gkronber
Message:

#2796 made several changes for debugging

Location:
branches/MCTS-SymbReg-2796
Files:
2 added
7 edited

Legend:

Unmodified
Added
Removed

  • branches/MCTS-SymbReg-2796/HeuristicLab.Algorithms.DataAnalysis/3.4/HeuristicLab.Algorithms.DataAnalysis.MCTSSymbReg.csproj

    r15416 r15425  
    9898  </ItemGroup>
    9999  <ItemGroup>
     100    <Compile Include="Heuristics.cs" />
    100101    <Compile Include="MctsSymbolicRegression\ApproximateDoubleEqualityComparer.cs" />
    101102    <Compile Include="MctsSymbolicRegression\IConstraintHandler.cs" />

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

    r15420 r15425  
    5656    //   
    5757
    58     // TODO: Taking averages of R² values is probably not ideal as an improvement of R² from 0.99 to 0.999 should
    59     //       weight more than an improvement from 0.98 to 0.99. Also, we are more interested in the best value of a
    60     //       branch and less in the expected value. (--> Review "Extreme Bandit" literature again)
     58    // TODO: The samples of x1*... or x2*... do not give any information about the relevance of the interaction term x1*x2 in general!
     59    //       --> E.g. if x1, x2 ~ N(0, 1) or U(-1, 1) this is trivial to show
     60    //       --> Therefore, looking at rollout statistics for arm selection is useless in the general case!
     61    //       --> It is necessary to rely on other features for the arm selection.
     62    //       --> TODO: Which heuristics can we apply?
    6163    // TODO: Solve Poly-10
    6264    // TODO: After state unification the recursive backpropagation of results takes a lot of time. How can this be improved?
    … …  
    7274    // TODO: improve memory usage
    7375    // TODO: support empty test partition
     76    // TODO: the algorithm should be invariant to linear transformations of the space (y = f(x') = f( Ax ) ) for invertible transformations A --> unit tests
    7477    #region static API
    7578
    … …  
    143146      // calculate for each level the number of alternatives the average 'inequality' of tries and 'inequality' of quality over the alternatives for each trie
    144147      // inequality can be calculated using the Gini coefficient
    145       internal readonly double[] giniCoeffs = new double[100];
    146      
     148      internal readonly double[] pathGiniCoeffs = new double[100];
     149      internal readonly double[] pathQs = new double[100];
     150      internal readonly double[] levelBestQ = new double[100];
     151      // internal readonly double[] levelMaxTries = new double[100];
     152      internal readonly double[] pathBestQ = new double[100]; // as long as pathBestQs = levelBestQs we are following the correct path
     153      internal readonly string[] levelBestAction = new string[100];
     154      internal readonly string[] curAction = new string[100];
     155      internal readonly double[] pathSelectedQ = new double[100];
    147156
    148157      public State(IRegressionProblemData problemData, uint randSeed, int maxVariables, bool scaleVariables,
    … …  
    448457#if DEBUG
    449458      internal void ClearStats() {
    450         for (int i = 0; i < giniCoeffs.Length; i++) giniCoeffs[i] = -1;
    451       }
    452       internal void WriteStats() {
    453         Console.WriteLine(string.Join("\t", giniCoeffs.TakeWhile(x => x >= 0).Select(x => string.Format("{0:N3}", x))));
    454       }
     459        for (int i = 0; i < pathGiniCoeffs.Length; i++) pathGiniCoeffs[i] = -1;
     460        for (int i = 0; i < pathQs.Length; i++) pathGiniCoeffs[i] = -99;
     461        for (int i = 0; i < pathBestQ.Length; i++) pathBestQ[i] = -99;
     462        for (int i = 0; i < pathSelectedQ.Length; i++) pathSelectedQ[i] = -99;
     463      }
     464      internal void WriteGiniStats() {
     465        Console.WriteLine(string.Join("\t", pathGiniCoeffs.TakeWhile(x => x >= 0).Select(x => string.Format("{0:N3}", x))));
     466      }
     467      internal void WriteQs() {
     468        // Console.WriteLine(string.Join("\t", pathQs.TakeWhile(x => x >= -100).Select(x => string.Format("{0:N3}", x))));
     469        var sb = new StringBuilder();
     470        // length
     471        int i = 0;
     472        while (i < pathBestQ.Length && pathBestQ[i] > -99 && pathBestQ[i] == levelBestQ[i]) {
     473          i++;
     474        }
     475        sb.AppendFormat("{0,-3}",i);
     476
     477        i = 0;
     478        // sb.AppendFormat("{0:N3}", levelBestQ[0]);
     479        while (i < pathSelectedQ.Length && pathSelectedQ[i] > -99) {
     480          sb.AppendFormat("\t{0:N3}", pathSelectedQ[i]);
     481          i++;
     482        }
     483        Console.WriteLine(sb.ToString());
     484        sb.Clear();
     485        i = 0;
     486        // sb.AppendFormat("{0:N3}", levelBestQ[0]);
     487        while (i < pathBestQ.Length && pathBestQ[i] > -99) {
     488          sb.AppendFormat("\t{0:N3}", pathBestQ[i]);
     489          i++;
     490        }
     491        Console.WriteLine(sb.ToString());
     492        sb.Clear();
     493        i = 0;
     494        while (i < pathBestQ.Length && pathBestQ[i] > -99) {
     495          sb.AppendFormat("\t{0:N3}", levelBestQ[i]);
     496          i++;
     497        }
     498        Console.WriteLine(sb.ToString());
     499       
     500        sb.Clear();
     501        i = 0;
     502        while (i < pathBestQ.Length && pathBestQ[i] > -99) {
     503          sb.AppendFormat("\t{0,-5}", (curAction[i] != null && curAction[i].Length > 5) ? curAction[i].Substring(0, 5) : curAction[i]);
     504          i++;
     505        }
     506        Console.WriteLine(sb.ToString());
     507        sb.Clear();
     508        i = 0;
     509        while (i < pathBestQ.Length && pathBestQ[i] > -99) {
     510          sb.AppendFormat("\t{0,-5}", (levelBestAction[i] != null && levelBestAction[i].Length > 5) ? levelBestAction[i].Substring(0, 5) : levelBestAction[i]);
     511          i++;
     512        }
     513        Console.WriteLine(sb.ToString());
     514       
     515        Console.WriteLine();
     516      }
     517
    455518
    456519#endif
    … …  
    525588
    526589#if DEBUG
    527       mctsState.WriteStats();
     590      // mctsState.WriteGiniStats();
     591      Console.WriteLine(ExprStr(automaton));
     592      mctsState.WriteQs();
     593      // Console.WriteLine(WriteStatistics(tree, mctsState));
     594
    528595#endif
    529596      //if (mctsState.effectiveRollouts % 100 == 1) {
    … …  
    564631            selectedIdx = treePolicy.Select(state.children[tree].Select(ch => ch.actionStatistics), rand);
    565632          }
    566 
     633         
    567634          // STATS
    568           state.giniCoeffs[tree.level] = InequalityCoefficient(state.children[tree].Select(ch => (double)ch.actionStatistics.AverageQuality));
     635          state.pathGiniCoeffs[tree.level] = InequalityCoefficient(state.children[tree].Select(ch => (double)ch.actionStatistics.AverageQuality));
     636          state.pathQs[tree.level] = tree.actionStatistics.AverageQuality;
    569637
    570638          tree = state.children[tree][selectedIdx];
    … …  
    576644          int nFs;
    577645          automaton.FollowStates(automaton.CurrentState, out possibleFollowStates, out nFs);
    578           while (automaton.CurrentState != tree.state && nFs == 1 &&
    579             !automaton.IsEvalState(possibleFollowStates[0]) && !automaton.IsFinalState(possibleFollowStates[0])) {
    580             automaton.Goto(possibleFollowStates[0]);
    581             automaton.FollowStates(automaton.CurrentState, out possibleFollowStates, out nFs);
    582           }
     646          // TODO!
     647          // while (possibleFollowStates[0] != tree.state && nFs == 1 &&
     648          //   !automaton.IsEvalState(possibleFollowStates[0]) && !automaton.IsFinalState(possibleFollowStates[0])) {
     649          //   automaton.Goto(possibleFollowStates[0]);
     650          //   automaton.FollowStates(automaton.CurrentState, out possibleFollowStates, out nFs);
     651          // }
    583652          Debug.Assert(possibleFollowStates.Contains(tree.state));
    584653          automaton.Goto(tree.state);
    … …  
    589658          string actionString = "";
    590659          automaton.FollowStates(automaton.CurrentState, out possibleFollowStates, out nFs);
    591           while (nFs == 1 && !automaton.IsEvalState(possibleFollowStates[0]) && !automaton.IsFinalState(possibleFollowStates[0])) {
    592             actionString += " " + automaton.GetActionString(automaton.CurrentState, possibleFollowStates[0]);
    593             // no alternatives -> just go to the next state
    594             automaton.Goto(possibleFollowStates[0]);
    595             automaton.FollowStates(automaton.CurrentState, out possibleFollowStates, out nFs);
    596           }
     660          // TODO
     661          // while (nFs == 1 && !automaton.IsEvalState(possibleFollowStates[0]) && !automaton.IsFinalState(possibleFollowStates[0])) {
     662          //   actionString += " " + automaton.GetActionString(automaton.CurrentState, possibleFollowStates[0]);
     663          //   // no alternatives -> just go to the next state
     664          //   automaton.Goto(possibleFollowStates[0]);
     665          //   automaton.FollowStates(automaton.CurrentState, out possibleFollowStates, out nFs);
     666          // }
    597667          if (nFs == 0) {
    598668            // stuck in a dead end (no final state and no allowed follow states)
    … …  
    672742        q = TransformQuality(q);
    673743        success = true;
     744        BackpropagateQuality(tree, q, treePolicy, state);       
    674745      } else {
    675746        // we got stuck in roll-out (not evaluation necessary!)
    … …  
    682753      // Update statistics
    683754      // Set branch to done if all children are done.
    684       BackpropagateQuality(tree, q, treePolicy, state);
     755      BackpropagateDone(tree, state);
     756      BackpropagateDebugStats(tree, q, state);
     757
    685758
    686759      return success;
    … …  
    703776    private static double TransformQuality(double q) {
    704777      // no transformation
    705       // return q;
     778      return q;
    706779
    707780      // EXPERIMENTAL!
    … …  
    738811
    739812    private static void BackpropagateQuality(Tree tree, double q, IPolicy policy, State state) {
    740       if (q > 0) policy.Update(tree.actionStatistics, q);
     813      policy.Update(tree.actionStatistics, q);
     814
     815      if (state.parents.ContainsKey(tree)) {
     816        foreach (var parent in state.parents[tree]) {
     817          BackpropagateQuality(parent, q, policy, state);
     818        }
     819      }
     820    }
     821
     822    private static void BackpropagateDone(Tree tree, State state) {
    741823      if (state.children.ContainsKey(tree) && state.children[tree].All(ch => ch.Done)) {
    742824        tree.Done = true;
    … …  
    746828      if (state.parents.ContainsKey(tree)) {
    747829        foreach (var parent in state.parents[tree]) {
    748           BackpropagateQuality(parent, q, policy, state);
    749         }
     830          BackpropagateDone(parent, state);
     831        }
     832      }
     833    }
     834
     835    private static void BackpropagateDebugStats(Tree tree, double q, State state) {
     836      if (state.parents.ContainsKey(tree)) {
     837        foreach (var parent in state.parents[tree]) {
     838          BackpropagateDebugStats(parent, q, state);
     839        }
     840      }
     841
     842      state.pathSelectedQ[tree.level] = tree.actionStatistics.AverageQuality;
     843      state.pathBestQ[tree.level] = tree.actionStatistics.BestQuality;
     844      state.curAction[tree.level] = tree.expr;
     845      if (state.levelBestQ[tree.level] < tree.actionStatistics.BestQuality) {
     846        state.levelBestQ[tree.level] = tree.actionStatistics.BestQuality;
     847        state.levelBestAction[tree.level] = tree.expr;
    750848      }
    751849    }
    … …  
    9171015    private static string WriteStatistics(Tree tree, State state) {
    9181016      var sb = new System.IO.StringWriter();
    919       sb.WriteLine("{0} {1:N5}", tree.actionStatistics.Tries, tree.actionStatistics.AverageQuality);
     1017      sb.Write("{0}\t{1:N5}\t", tree.actionStatistics.Tries, tree.actionStatistics.AverageQuality);
    9201018      if (state.children.ContainsKey(tree)) {
    9211019        foreach (var ch in state.children[tree]) {
    922           sb.WriteLine("{0} {1:N5}", ch.actionStatistics.Tries, ch.actionStatistics.AverageQuality);
    923         }
    924       }
     1020          sb.Write("{0}\t{1:N5}\t", ch.actionStatistics.Tries, ch.actionStatistics.AverageQuality);
     1021        }
     1022      }
     1023      sb.WriteLine();
    9251024      return sb.ToString();
    9261025    }

  • branches/MCTS-SymbReg-2796/HeuristicLab.Algorithms.DataAnalysis/3.4/MctsSymbolicRegression/Policies/EpsGreedy.cs

    r15410 r15425  
    1818      public double SumQuality { get; set; }
    1919      public double AverageQuality { get { return SumQuality / Tries; } }
     20      public double BestQuality { get; internal set; }
    2021      public int Tries { get; set; }
    2122      public bool Done { get; set; }
    … …  
    2627        this.Tries += o.Tries;
    2728        this.SumQuality += o.SumQuality;
     29        this.BestQuality = Math.Max(this.BestQuality, other.BestQuality);
    2830      }
    2931    }
    … …  
    6062      var a = action as ActionStatistics;
    6163      a.SumQuality += q;
     64      a.BestQuality = Math.Max(a.BestQuality, q);
    6265      a.Tries++;
    6366    }

  • branches/MCTS-SymbReg-2796/HeuristicLab.Algorithms.DataAnalysis/3.4/MctsSymbolicRegression/Policies/IActionStatistics.cs

    r15410 r15425  
    88  public interface IActionStatistics {
    99    double AverageQuality { get; }
     10    double BestQuality { get; }
    1011    int Tries { get; }
    1112    bool Done { get; set; }

  • branches/MCTS-SymbReg-2796/HeuristicLab.Algorithms.DataAnalysis/3.4/MctsSymbolicRegression/Policies/Ucb.cs

    r15416 r15425  
    1919      public double SumQuality { get; set; }
    2020      public double AverageQuality { get { return SumQuality / Tries; } }
     21      public double BestQuality { get; internal set; }
    2122      public int Tries { get; set; }
    2223      public bool Done { get; set; }
    … …  
    2627        this.Tries += o.Tries;
    2728        this.SumQuality += o.SumQuality;
     29        this.BestQuality = Math.Max(this.BestQuality, other.BestQuality);
    2830      }
    2931    }
    … …  
    6062      var a = action as ActionStatistics;
    6163      a.SumQuality += q;
     64      a.BestQuality = Math.Max(a.BestQuality, q);
    6265      a.Tries++;
    6366    }
    … …  
    8285        return buf[rand.Next(buf.Count)];
    8386      }
     87
     88      Debug.Assert(actions.All(a => a.Done || a.Tries > 0));
    8489                     
    8590      Debug.Assert(totalTries > 0);

  • branches/MCTS-SymbReg-2796/HeuristicLab.Algorithms.DataAnalysis/3.4/MctsSymbolicRegression/Policies/UcbTuned.cs

    r15410 r15425  
    1919      public double SumSqrQuality { get; set; }
    2020      public double AverageQuality { get { return SumQuality / Tries; } }
     21      public double BestQuality { get; internal set; }
    2122      public double QualityVariance { get { return SumSqrQuality / Tries - AverageQuality * AverageQuality; } }
    2223      public int Tries { get; set; }
    … …  
    2930        this.SumQuality += o.SumQuality;
    3031        this.SumSqrQuality += o.SumSqrQuality;
     32        this.BestQuality = Math.Max(this.BestQuality, other.BestQuality);
    3133      }
    3234    }
    … …  
    6466      a.SumQuality += q;
    6567      a.SumSqrQuality += q * q;
     68      a.BestQuality = Math.Max(a.BestQuality, q);
    6669      a.Tries++;
    6770    }

  • branches/MCTS-SymbReg-2796/Tests/HeuristicLab.Algorithms.DataAnalysis-3.4/MctsSymbolicRegressionTest.cs

    r15420 r15425  
    44using System.Threading;
    55using HeuristicLab.Algorithms.DataAnalysis.MctsSymbolicRegression.Policies;
     6using HeuristicLab.Algorithms.DataAnalysis.MCTSSymbReg;
    67using HeuristicLab.Data;
    78using HeuristicLab.Optimization;
    … …  
    1415  [TestClass()]
    1516  public class MctsSymbolicRegressionTest {
     17    #region heuristics
     18    [TestMethod]
     19    [TestCategory("Algorithms.DataAnalysis")]
     20    [TestProperty("Time", "short")]
     21    public void TestHeuristics() {
     22      {
     23        // a, b ~ U(0, 1) should be trivial
     24        var nRand = new MersenneTwister(1234);
     25
     26        int n = 10000; // large sample so that we can use the thresholds below
     27        var a = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     28        var b = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     29        var x = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     30        var y = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     31
     32        var z = a.Zip(b, (ai, bi) => ai * bi).ToArray();
     33
     34        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, b, z) > 0.05); // should be detected as relevant
     35        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, x, z) > 0.05); // a and b > 0 so these should be detected as well
     36        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, y, z) > 0.05);
     37        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, x, z) > 0.05);
     38        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, y, z) > 0.05);
     39        Assert.IsTrue(Heuristics.CorrelationForInteraction(x, y, z) < 0.05);
     40      }
     41      {
     42        // a, b ~ U(1000, 2000) also trivial
     43        var nRand = new UniformDistributedRandom(new MersenneTwister(1234), 1000, 2000);
     44
     45        int n = 10000; // large sample so that we can use the thresholds below
     46        var a = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     47        var b = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     48        var x = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     49        var y = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     50
     51        var z = a.Zip(b, (ai, bi) => ai * bi).ToArray();
     52
     53        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, b, z) > 0.05); // should be detected as relevant
     54        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, x, z) > 0.05);
     55        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, y, z) > 0.05);
     56        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, x, z) > 0.05);
     57        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, y, z) > 0.05);
     58        Assert.IsTrue(Heuristics.CorrelationForInteraction(x, y, z) < 0.05);
     59      }
     60      {
     61        // a, b ~ U(-1, 1)
     62        var nRand = new UniformDistributedRandom(new MersenneTwister(1234), -1, 1);
     63
     64        int n = 10000; // large sample so that we can use the thresholds below
     65        var a = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     66        var b = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     67        var x = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     68        var y = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     69
     70        var z = a.Zip(b, (ai, bi) => ai * bi).ToArray();
     71
     72        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, b, z) > 0.05); // should be detected as relevant
     73        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, x, z) < 0.05);
     74        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, y, z) < 0.05);
     75        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, x, z) < 0.05);
     76        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, y, z) < 0.05);
     77        Assert.IsTrue(Heuristics.CorrelationForInteraction(x, y, z) < 0.05);
     78      }
     79      {
     80        // a, b ~ N(0, 1)
     81        var nRand = new NormalDistributedRandom(new MersenneTwister(1234), 0, 1);
     82
     83        int n = 10000; // large sample so that we can use the thresholds below
     84        var a = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     85        var b = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     86        var x = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     87        var y = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     88
     89        var z = a.Zip(b, (ai, bi) => ai * bi).ToArray();
     90
     91        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, b, z) > 0.05); // should be detected as relevant
     92        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, x, z) < 0.05);
     93        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, y, z) < 0.05);
     94        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, x, z) < 0.05);
     95        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, y, z) < 0.05);
     96        Assert.IsTrue(Heuristics.CorrelationForInteraction(x, y, z) < 0.05);
     97      }
     98      {
     99        // a ~ N(100, 1), b ~ N(-100, 1)
     100        var nRand = new NormalDistributedRandom(new MersenneTwister(1234), 0, 1);
     101        var aRand = new NormalDistributedRandom(new MersenneTwister(1234), 100, 1);
     102        var bRand = new NormalDistributedRandom(new MersenneTwister(1234), -100, 1);
     103
     104        int n = 10000; // large sample so that we can use the thresholds below
     105        var a = Enumerable.Range(0, n).Select(_ => aRand.NextDouble()).ToArray();
     106        var b = Enumerable.Range(0, n).Select(_ => bRand.NextDouble()).ToArray();
     107        var x = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     108        var y = Enumerable.Range(0, n).Select(_ => nRand.NextDouble()).ToArray();
     109
     110        var z = a.Zip(b, (ai, bi) => ai * bi).ToArray();
     111
     112        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, b, z) > 0.05); // should be detected as relevant
     113        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, x, z) > 0.05); // a > 0
     114        Assert.IsTrue(Heuristics.CorrelationForInteraction(a, y, z) > 0.05);
     115        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, x, z) > 0.05); // b < 0
     116        Assert.IsTrue(Heuristics.CorrelationForInteraction(b, y, z) > 0.05);
     117        Assert.IsTrue(Heuristics.CorrelationForInteraction(x, y, z) < 0.05);
     118      }
     119    }
     120    #endregion
     121
     122
    16123    #region expression hashing
    17124    [TestMethod]
    … …  
    790897      }
    791898
    792       var ds = new Dataset(new string[] { "a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "y" }, 
     899      var ds = new Dataset(new string[] { "a", "b", "c", "d", "e", "f", "g", "h", "i", "j", "y" },
    793900        new[] { x1, x2, x3, x4, x5, x6, x7, x8, x9, x10, ys });
    794901
    795902
    796       var problemData = new RegressionProblemData(ds, new string[] { "a", "b", "c", "d", "e", "f", "g", "h", "i", "j"}, "y");
     903      var problemData = new RegressionProblemData(ds, new string[] { "a", "b", "c", "d", "e", "f", "g", "h", "i", "j" }, "y");
    797904
    798905      problemData.TrainingPartition.Start = 0;
    … …  
    11181225
    11191226      // UCB tuned
    1120       // var ucbTuned = new UcbTuned();
    1121       // ucbTuned.C = 1.5;
    1122       // mctsSymbReg.Policy = ucbTuned;
     1227      var ucbTuned = new UcbTuned();
     1228      ucbTuned.C = 1;
     1229      mctsSymbReg.Policy = ucbTuned;
    11231230
    11241231
Note: See TracChangeset for help on using the changeset viewer.