public static void SupervisedKernelHgForLabel(int Rows, LabeledImage[] images, int InLabel)
{
// Select training set and test set.
const int TrainingSize = 9000;
int TestSize = Rows - TrainingSize;
var TrainingSet = Subset(Rows, images, TrainingSize);
var TestSet = Remainder(images, TrainingSet);
// Select smaller set to generate leads with.
const int LeadSize = 200;
const int InLabelSize = LeadSize / 2;
LabeledImage[] LeadSet = SampleSet(TrainingSize, TrainingSet, InLabel, LeadSize, InLabelSize);
// Get list of candidate kernels from the lead set.
var list = Learn_SupervisedKernelHg(LeadSize, LeadSet, InLabel: InLabel, InLabelMinRatio: 0.2f, SearchLength: 50);
list.Sort((kernelScore1, kernelScore2) => kernelScore2.Item2.CompareTo(kernelScore1.Item2));
foreach (var kernelScore in list)
{
Console.WriteLine($"{kernelScore.Item2}");
}
// Test candidates on test set.
const int BlockDim = 5;
var blocks = GetBlocks(TestSet, TestSize, Width, Height, Channels: 3, BlockDim: BlockDim, AddMirrors: true);
foreach (var kernel in list)
{
var tracker = new BlockTracker();
var result = GetScoreStats(TestSize, blocks, kernel.Item1, InLabel, tracker);
Console.WriteLine($"Score is {result.InOutHitRatio} with an in ratio of {result.InHitRatio}");
//pl.plot(Average(tracker.in_remainders), Average(tracker.out_remainders));
//pl.plot(Average(tracker.in_blocks), Average(tracker.out_blocks));
pl.plot(minus(Average(tracker.in_remainders), Average(tracker.out_remainders)));
pl.plot(minus(Average(tracker.in_blocks), Average(tracker.out_blocks)));
//var score = OptimizeKernelCutoff(TestSize, blocks, kernel.Item1, InLabel: InLabel, InLabelMinRatio: 0.2f);
//Console.WriteLine($"Score is {score.InOutHitRatio} with an in ratio of {score.InHitRatio}");
}
}
static ScoreStats GetScoreStats(int Rows, BlockedImage[] images, Kernel kernel, int InLabel, BlockTracker tracker = null)
{
var stats = new ScoreStats();
for (int i = 0; i < Rows; i++)
{
var image = images[i];
var action = tracker == null ? null : tracker.GetTrackInOut(image, InLabel);
float error = GetScore(image, kernel, action: action);
if (image.Label == InLabel)
{
stats.InCount++;
if (error < kernel.ClassificationCutoff)
{
stats.InHit++;
}
}
else
{
stats.OutCount++;
if (error < kernel.ClassificationCutoff)
{
stats.OutHit++;
}
}
}
Console.WriteLine($" - in label {stats.InHit} / {stats.InCount} - out label {stats.OutHit} / {stats.OutCount} - {kernel.ShortDescription()}");
return(stats);
}