public void SoftmaxScores()
{
int numActions = 10;
float lambda = 0.5f;
var recorder = new TestRecorder <RegularTestContext>();
var scorer = new TestScorer <RegularTestContext>(1, numActions, uniform: false);
//var mwtt = new MwtExplorer<RegularTestContext>("mwt", recorder);
var explorer = new SoftmaxExplorer(lambda);
var mwtt = MwtExplorer.Create("mwt", numActions, recorder, explorer, scorer);
Random rand = new Random();
mwtt.ChooseAction(rand.NextDouble().ToString(), new RegularTestContext()
{
Id = 100
});
mwtt.ChooseAction(rand.NextDouble().ToString(), new RegularTestContext()
{
Id = 101
});
mwtt.ChooseAction(rand.NextDouble().ToString(), new RegularTestContext()
{
Id = 102
});
var interactions = recorder.GetAllInteractions();
Assert.AreEqual(3, interactions.Count);
for (int i = 0; i < interactions.Count; i++)
{
// Scores are not equal therefore probabilities should not be uniform
Assert.AreNotEqual(interactions[i].Probability, 1.0f / numActions);
Assert.AreEqual(100 + i, interactions[i].Context.Id);
}
// Verify that policy action is chosen all the time
RegularTestContext context = new RegularTestContext {
Id = 100
};
List <float> scores = scorer.MapContext(context).Value.ToList();
float maxScore = 0;
int highestScoreAction = 0;
for (int i = 0; i < scores.Count; i++)
{
if (maxScore < scores[i])
{
maxScore = scores[i];
highestScoreAction = i + 1;
}
}
explorer.EnableExplore(false);
for (int i = 0; i < 1000; i++)
{
int chosenAction = mwtt.ChooseAction(rand.NextDouble().ToString(), new RegularTestContext()
{
Id = (int)i
});
Assert.AreEqual(highestScoreAction, chosenAction);
}
}
