1 | using System;
|
---|
2 | using System.Collections.Generic;
|
---|
3 | using System.Diagnostics;
|
---|
4 | using System.Linq;
|
---|
5 | using System.Text;
|
---|
6 | using System.Threading.Tasks;
|
---|
7 |
|
---|
8 | namespace HeuristicLab.Algorithms.Bandits {
|
---|
9 | public class UCB1TunedPolicy : BanditPolicy {
|
---|
10 | private readonly int[] tries;
|
---|
11 | private readonly double[] sumReward;
|
---|
12 | private readonly double[] sumSqrReward;
|
---|
13 | private int totalTries = 0;
|
---|
14 | public UCB1TunedPolicy(int numActions)
|
---|
15 | : base(numActions) {
|
---|
16 | this.tries = new int[numActions];
|
---|
17 | this.sumReward = new double[numActions];
|
---|
18 | this.sumSqrReward = new double[numActions];
|
---|
19 | }
|
---|
20 |
|
---|
21 | private double V(int arm) {
|
---|
22 | var s = tries[arm];
|
---|
23 | return sumSqrReward[arm] / s - Math.Pow(sumReward[arm] / s, 2) + Math.Sqrt(2 * Math.Log(totalTries) / s);
|
---|
24 | }
|
---|
25 |
|
---|
26 |
|
---|
27 | public override int SelectAction() {
|
---|
28 | Debug.Assert(Actions.Any());
|
---|
29 | int bestAction = -1;
|
---|
30 | double bestQ = double.NegativeInfinity;
|
---|
31 | foreach (var a in Actions) {
|
---|
32 | if (tries[a] == 0) return a;
|
---|
33 | var q = sumReward[a] / tries[a] + Math.Sqrt((Math.Log(totalTries) / tries[a]) * Math.Min(1.0 / 4, V(a))); // 1/4 is upper bound of bernoulli distributed variable
|
---|
34 | if (q > bestQ) {
|
---|
35 | bestQ = q;
|
---|
36 | bestAction = a;
|
---|
37 | }
|
---|
38 | }
|
---|
39 | return bestAction;
|
---|
40 | }
|
---|
41 | public override void UpdateReward(int action, double reward) {
|
---|
42 | Debug.Assert(Actions.Contains(action));
|
---|
43 | totalTries++;
|
---|
44 | tries[action]++;
|
---|
45 | sumReward[action] += reward;
|
---|
46 | sumSqrReward[action] += reward * reward;
|
---|
47 | }
|
---|
48 |
|
---|
49 | public override void DisableAction(int action) {
|
---|
50 | base.DisableAction(action);
|
---|
51 | totalTries -= tries[action];
|
---|
52 | tries[action] = -1;
|
---|
53 | sumReward[action] = 0;
|
---|
54 | sumSqrReward[action] = 0;
|
---|
55 | }
|
---|
56 |
|
---|
57 | public override void Reset() {
|
---|
58 | base.Reset();
|
---|
59 | totalTries = 0;
|
---|
60 | Array.Clear(tries, 0, tries.Length);
|
---|
61 | Array.Clear(sumReward, 0, sumReward.Length);
|
---|
62 | Array.Clear(sumSqrReward, 0, sumSqrReward.Length);
|
---|
63 | }
|
---|
64 | }
|
---|
65 | }
|
---|