Free cookie consent management tool by TermsFeed Policy Generator

# source:branches/HeuristicLab.Problems.GrammaticalOptimization/HeuristicLab.Algorithms.Bandits/Policies/UCB1Policy.cs@11728

Last change on this file since 11728 was 11727, checked in by gkronber, 8 years ago

#2283: worked on grammatical optimization problem solvers (simple MCTS done)

File size: 1.5 KB
RevLine
[11710]1using System;
2using System.Collections.Generic;
[11727]3using System.Diagnostics;
[11710]4using System.Linq;
5using System.Text;
7
8namespace HeuristicLab.Algorithms.Bandits {
9  public class UCB1Policy : BanditPolicy {
12    private int totalTries = 0;
13    public UCB1Policy(int numActions)
14      : base(numActions) {
[11727]15      this.tries = new int[numActions];
16      this.sumReward = new double[numActions];
[11710]17    }
18
19    public override int SelectAction() {
20      int bestAction = -1;
21      double bestQ = double.NegativeInfinity;
[11727]22      foreach (var a in Actions) {
23        if (tries[a] == 0) return a;
24        var q = sumReward[a] / tries[a] + Math.Sqrt((2 * Math.Log(totalTries)) / tries[a]);
[11710]25        if (q > bestQ) {
26          bestQ = q;
[11727]27          bestAction = a;
[11710]28        }
29      }
30      return bestAction;
31    }
32    public override void UpdateReward(int action, double reward) {
[11727]33      Debug.Assert(Actions.Contains(action));
[11710]34      totalTries++;
35      tries[action]++;
36      sumReward[action] += reward;
37    }
[11727]38
39    public override void DisableAction(int action) {
40      base.DisableAction(action);
41      totalTries -= tries[action];
42      tries[action] = -1;
43      sumReward[action] = 0;
44    }
45
[11710]46    public override void Reset() {
[11727]47      base.Reset();
[11710]48      totalTries = 0;
49      Array.Clear(tries, 0, tries.Length);
50      Array.Clear(sumReward, 0, sumReward.Length);
51    }
52  }
53}
Note: See TracBrowser for help on using the repository browser.