apaam/mlpack
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Don't use deprecated functions in tests; use new versions.

Browse Source
This commit is contained in:
Ryan Curtin
2023-11-13 16:55:20 -05:00
parent 1e5375edb1
commit 27b9bb335e
1 changed files with 55 additions and 87 deletions
Download Patch File Download Diff File Expand all files Collapse all files
src/mlpack/tests/adaboost_test.cpp
+55 -87
View File
@@ -19,8 +19,6 @@
using namespace arma;
using namespace mlpack;
// TODO: adapt to non-deprecated calls
/**
* This test case runs the AdaBoost.mh algorithm on the UCI Iris dataset. It
* checks whether the hamming loss breaches the upperbound, which is provided by
@@ -46,16 +44,14 @@ TEMPLATE_TEST_CASE("HammingLossBoundIris", "[AdaBoostTest]", mat, fmat)
// Run the perceptron for perceptronIter iterations.
int perceptronIter = 400;
typedef Perceptron<SimpleWeightUpdate, ZeroInitialization, MatType>
PerceptronType;
PerceptronType p(inputData, labels.row(0), numClasses, perceptronIter);
// Define parameters for AdaBoost.
size_t iterations = 100;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(tolerance);
eT ztProduct = a.Train(inputData, labels.row(0), numClasses, p, iterations,
tolerance);
eT tolerance = 2e-10;
typedef Perceptron<SimpleWeightUpdate, ZeroInitialization, MatType>
PerceptronType;
AdaBoost<PerceptronType, MatType> a;
eT ztProduct = a.Train(inputData, labels.row(0), numClasses, iterations,
tolerance, perceptronIter);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -79,12 +75,10 @@ TEMPLATE_TEST_CASE("WeakLearnerErrorIris", "[AdaBoostTest]", mat, fmat)
typedef typename MatType::elem_type eT;
MatType inputData;
if (!data::Load("iris.csv", inputData))
FAIL("Cannot load test dataset iris.csv!");
Mat<size_t> labels;
if (!data::Load("iris_labels.txt", labels))
FAIL("Cannot load labels for iris iris_labels.txt");
@@ -106,13 +100,13 @@ TEMPLATE_TEST_CASE("WeakLearnerErrorIris", "[AdaBoostTest]", mat, fmat)
// Define parameters for AdaBoost.
size_t iterations = 100;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses, p,
iterations, tolerance);
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses,
iterations, tolerance, perceptronIter);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
size_t countError = accu(labels != predictedLabels);;
size_t countError = accu(labels != predictedLabels);
eT error = (eT) countError / labels.n_cols;
REQUIRE(error <= weakLearnerErrorRate + 0.03);
@@ -150,8 +144,8 @@ TEMPLATE_TEST_CASE("HammingLossBoundVertebralColumn", "[AdaBoostTest]", mat,
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(tolerance);
eT ztProduct = a.Train(inputData, labels.row(0), numClasses, p, iterations,
tolerance);
eT ztProduct = a.Train(inputData, labels.row(0), numClasses, iterations,
tolerance, perceptronIter);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -201,8 +195,8 @@ TEMPLATE_TEST_CASE("WeakLearnerErrorVertebralColumn", "[AdaBoostTest]", mat,
// Define parameters for AdaBoost.
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses, p,
iterations, tolerance);
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses,
iterations, tolerance, perceptronIter);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -245,8 +239,8 @@ TEMPLATE_TEST_CASE("HammingLossBoundNonLinearSepData", "[AdaBoostTest]", mat,
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(tolerance);
eT ztProduct = a.Train(inputData, labels.row(0), numClasses, p, iterations,
tolerance);
eT ztProduct = a.Train(inputData, labels.row(0), numClasses, iterations,
tolerance, perceptronIter);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -296,8 +290,8 @@ TEMPLATE_TEST_CASE("WeakLearnerErrorNonLinearSepData", "[AdaBoostTest]", mat,
// Define parameters for AdaBoost.
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses, p,
iterations, tolerance);
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses,
iterations, tolerance, perceptronIter);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -330,14 +324,13 @@ TEMPLATE_TEST_CASE("HammingLossIris_DS", "[AdaBoostTest]", mat, fmat)
const size_t numClasses = 3;
const size_t inpBucketSize = 6;
Row<size_t> labelsvec = labels.row(0);
ID3DecisionStump ds(inputData, labelsvec, numClasses, inpBucketSize);
// Define parameters for AdaBoost.
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<ID3DecisionStump, MatType> a(tolerance);
eT ztProduct = a.Train(inputData, labelsvec, numClasses, ds, iterations,
tolerance);
eT ztProduct = a.Train(inputData, labelsvec, numClasses, iterations,
tolerance, inpBucketSize);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -388,8 +381,8 @@ TEMPLATE_TEST_CASE("WeakLearnerErrorIris_DS", "[AdaBoostTest]", mat, fmat)
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<ID3DecisionStump, MatType> a(inputData, labelsvec, numClasses, ds,
iterations, tolerance);
AdaBoost<ID3DecisionStump, MatType> a(inputData, labelsvec, numClasses,
iterations, tolerance, inpBucketSize);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -432,8 +425,8 @@ TEMPLATE_TEST_CASE("HammingLossBoundVertebralColumn_DS", "[AdaBoostTest]", mat,
eT tolerance = 1e-10;
AdaBoost<ID3DecisionStump, MatType> a(tolerance);
eT ztProduct = a.Train(inputData, labelsvec, numClasses, ds, iterations,
tolerance);
eT ztProduct = a.Train(inputData, labelsvec, numClasses, iterations,
tolerance, inpBucketSize);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -481,8 +474,8 @@ TEMPLATE_TEST_CASE("WeakLearnerErrorVertebralColumn_DS", "[AdaBoostTest]", mat,
// Define parameters for AdaBoost.
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<ID3DecisionStump, MatType> a(inputData, labelsvec, numClasses, ds,
iterations, tolerance);
AdaBoost<ID3DecisionStump, MatType> a(inputData, labelsvec, numClasses,
iterations, tolerance, inpBucketSize);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -524,8 +517,8 @@ TEMPLATE_TEST_CASE("HammingLossBoundNonLinearSepData_DS", "[AdaBoostTest]", mat,
eT tolerance = 1e-10;
AdaBoost<ID3DecisionStump, MatType> a(tolerance);
eT ztProduct = a.Train(inputData, labelsvec, numClasses, ds, iterations,
tolerance);
eT ztProduct = a.Train(inputData, labelsvec, numClasses, iterations,
tolerance, inpBucketSize);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -575,8 +568,8 @@ TEMPLATE_TEST_CASE("WeakLearnerErrorNonLinearSepData_DS", "[AdaBoostTest]", mat,
size_t iterations = 500;
eT tolerance = 1e-23;
AdaBoost<ID3DecisionStump, MatType> a(inputData, labelsvec, numClasses, ds,
iterations, tolerance);
AdaBoost<ID3DecisionStump, MatType> a(inputData, labelsvec, numClasses,
iterations, tolerance, inpBucketSize);
Row<size_t> predictedLabels;
a.Classify(inputData, predictedLabels);
@@ -629,8 +622,8 @@ TEMPLATE_TEST_CASE("ClassifyTest_VERTEBRALCOL", "[AdaBoostTest]", mat, fmat)
// Define parameters for AdaBoost.
size_t iterations = 100;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses, p,
iterations, tolerance);
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses,
iterations, tolerance, perceptronIter);
Row<size_t> predictedLabels1(testData.n_cols),
predictedLabels2(testData.n_cols);
@@ -700,8 +693,8 @@ TEMPLATE_TEST_CASE("ClassifyTest_NONLINSEP", "[AdaBoostTest]", mat, fmat)
// Define parameters for AdaBoost.
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<ID3DecisionStump, MatType> a(inputData, labelsvec, numClasses, ds,
iterations, tolerance);
AdaBoost<ID3DecisionStump, MatType> a(inputData, labelsvec, numClasses,
iterations, tolerance, inpBucketSize);
Row<size_t> predictedLabels1(testData.n_cols),
predictedLabels2(testData.n_cols);
@@ -762,8 +755,8 @@ TEMPLATE_TEST_CASE("ClassifyTest_IRIS", "[AdaBoostTest]", mat, fmat)
// Define parameters for AdaBoost.
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses, p,
iterations, tolerance);
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses,
iterations, tolerance, perceptronIter);
MatType testData;
if (!data::Load("iris_test.csv", testData))
@@ -832,8 +825,8 @@ TEMPLATE_TEST_CASE("TrainTest", "[AdaBoostTest]", mat, fmat)
// Now train AdaBoost.
size_t iterations = 50;
eT tolerance = 1e-10;
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses, p,
iterations, tolerance);
AdaBoost<PerceptronType, MatType> a(inputData, labels.row(0), numClasses,
iterations, tolerance, perceptronIter);
// Now load another dataset...
if (!data::Load("vc2.csv", inputData))
@@ -845,7 +838,8 @@ TEMPLATE_TEST_CASE("TrainTest", "[AdaBoostTest]", mat, fmat)
PerceptronType p2(inputData, labels.row(0), newNumClasses, perceptronIter);
a.Train(inputData, labels.row(0), newNumClasses, p2, iterations, tolerance);
a.Train(inputData, labels.row(0), newNumClasses, iterations, tolerance,
perceptronIter);
// Load test set to see if it trained on vc2 correctly.
MatType testData;
@@ -860,7 +854,7 @@ TEMPLATE_TEST_CASE("TrainTest", "[AdaBoostTest]", mat, fmat)
Row<size_t> predictedLabels(testData.n_cols);
a.Classify(testData, predictedLabels);
int localError = accu(trueTestLabels != predictedLabels);
size_t localError = accu(trueTestLabels != predictedLabels);
eT lError = (eT) localError / trueTestLabels.n_cols;
REQUIRE(lError <= 0.30);
@@ -880,8 +874,7 @@ TEMPLATE_TEST_CASE("PerceptronSerializationTest", "[AdaBoostTest]", fmat, mat)
typedef Perceptron<SimpleWeightUpdate, ZeroInitialization, MatType>
PerceptronType;
PerceptronType p(data, labels, 2, 800);
AdaBoost<PerceptronType, MatType> ab(data, labels, 2, p, 50, 1e-10);
AdaBoost<PerceptronType, MatType> ab(data, labels, 2, 50, 1e-10, 800);
// Now create another dataset to train with.
MatType otherData = randu<MatType>(5, 200);
@@ -893,9 +886,8 @@ TEMPLATE_TEST_CASE("PerceptronSerializationTest", "[AdaBoostTest]", fmat, mat)
for (size_t i = 150; i < 200; ++i)
otherLabels[i] = 2;
PerceptronType p2(otherData, otherLabels, 3, 500);
AdaBoost<PerceptronType, MatType> abText(otherData, otherLabels, 3, p2, 50,
1e-10);
AdaBoost<PerceptronType, MatType> abText(otherData, otherLabels, 3, 50, 1e-10,
500);
AdaBoost<PerceptronType, MatType> abXml, abBinary;
@@ -937,8 +929,7 @@ TEMPLATE_TEST_CASE("ID3DecisionStumpSerializationTest", "[AdaBoostTest]", mat,
for (size_t i = 250; i < 500; ++i)
labels[i] = 1;
ID3DecisionStump p(data, labels, 2, 800);
AdaBoost<ID3DecisionStump, MatType> ab(data, labels, 2, p, 50, 1e-10);
AdaBoost<ID3DecisionStump, MatType> ab(data, labels, 2, 50, 1e-10, 40);
// Now create another dataset to train with.
MatType otherData = randu<MatType>(5, 200);
@@ -950,9 +941,8 @@ TEMPLATE_TEST_CASE("ID3DecisionStumpSerializationTest", "[AdaBoostTest]", mat,
for (size_t i = 150; i < 200; ++i)
otherLabels[i] = 2;
ID3DecisionStump p2(otherData, otherLabels, 3, 500);
AdaBoost<ID3DecisionStump, MatType> abText(otherData, otherLabels, 3, p2, 50,
1e-10);
AdaBoost<ID3DecisionStump, MatType> abText(otherData, otherLabels, 3, 50,
1e-10, 25);
AdaBoost<ID3DecisionStump, MatType> abXml, abBinary;
@@ -970,7 +960,7 @@ TEMPLATE_TEST_CASE("ID3DecisionStumpSerializationTest", "[AdaBoostTest]", mat,
for (size_t i = 0; i < ab.WeakLearners(); ++i)
{
REQUIRE(ab.WeakLearner(i).SplitDimension() ==
abXml.WeakLearner(i).SplitDimension());
abXml.WeakLearner(i).SplitDimension());
REQUIRE(ab.WeakLearner(i).SplitDimension() ==
abText.WeakLearner(i).SplitDimension());
REQUIRE(ab.WeakLearner(i).SplitDimension() ==
@@ -1034,7 +1024,6 @@ TEMPLATE_TEST_CASE("AdaBoostSinglePointClassifyWithProbs", "[AdaBoostTest]",
TEMPLATE_TEST_CASE("AdaBoostParamsConstructor", "[AdaBoostTest]", fmat, mat)
{
typedef TestType MatType;
typedef typename MatType::elem_type ElemType;
MatType inputData;
if (!data::Load("iris.csv", inputData))
@@ -1073,7 +1062,6 @@ TEMPLATE_TEST_CASE("AdaBoostParamsConstructor", "[AdaBoostTest]", fmat, mat)
TEMPLATE_TEST_CASE("AdaBoostTrainOverloads", "[AdaBoostTest]", fmat, mat)
{
typedef TestType MatType;
typedef typename MatType::elem_type ElemType;
// Create random data.
MatType data = randu<MatType>(10, 100);
@@ -1082,49 +1070,32 @@ TEMPLATE_TEST_CASE("AdaBoostTrainOverloads", "[AdaBoostTest]", fmat, mat)
typedef Perceptron<SimpleWeightUpdate, ZeroInitialization, MatType>
PerceptronType;
PerceptronType p; // For versions that take an initialized weak learner.
p.MaxIterations() = 150;
AdaBoost<PerceptronType, MatType> a1, a2, a3, a4, a5, a6, a7;
a1.MaxIterations() = 75;
a4.MaxIterations() = 65;
a1.Tolerance() = 1e-4;
a2.Tolerance() = 1e-5;
a4.Tolerance() = 2e-4;
a5.Tolerance() = 2e-5;
AdaBoost<PerceptronType, MatType> a1, a2, a3, a4;
a1.MaxIterations() = 65;
a1.Tolerance() = 2e-4;
a2.Tolerance() = 2e-5;
a1.Train(data, labels, 4, p);
a2.Train(data, labels, 4, p, 10);
a3.Train(data, labels, 4, p, 50, 1e-3);
a4.Train(data, labels, 4);
a5.Train(data, labels, 4, 15);
a6.Train(data, labels, 4, 55, 1e-3);
a7.Train(data, labels, 4, 60, 2e-3, 100);
a1.Train(data, labels, 4);
a2.Train(data, labels, 4, 15);
a3.Train(data, labels, 4, 55, 1e-3);
a4.Train(data, labels, 4, 60, 2e-3, 100);
// Make sure hyperparameters were set correctly, where appropriate.
REQUIRE(a1.MaxIterations() == 75);
REQUIRE(a2.MaxIterations() == 10);
REQUIRE(a3.MaxIterations() == 50);
REQUIRE(a4.MaxIterations() == 65);
REQUIRE(a5.MaxIterations() == 15);
REQUIRE(a6.MaxIterations() == 55);
REQUIRE(a7.MaxIterations() == 60);
REQUIRE(a1.Tolerance() == Approx(1e-4));
REQUIRE(a2.Tolerance() == Approx(1e-5));
REQUIRE(a3.Tolerance() == Approx(1e-3));
REQUIRE(a4.Tolerance() == Approx(2e-4));
REQUIRE(a5.Tolerance() == Approx(2e-5));
REQUIRE(a6.Tolerance() == Approx(1e-3));
REQUIRE(a7.Tolerance() == Approx(2e-3));
// Make sure anything at all was trained.
REQUIRE(a1.WeakLearners() > 0);
REQUIRE(a2.WeakLearners() > 0);
REQUIRE(a3.WeakLearners() > 0);
REQUIRE(a4.WeakLearners() > 0);
REQUIRE(a5.WeakLearners() > 0);
REQUIRE(a6.WeakLearners() > 0);
REQUIRE(a7.WeakLearners() > 0);
// Make sure the maximum number of iterations in the perceptron was set
// properly.
@@ -1132,7 +1103,4 @@ TEMPLATE_TEST_CASE("AdaBoostTrainOverloads", "[AdaBoostTest]", fmat, mat)
REQUIRE(a2.WeakLearner(0).MaxIterations() == 150);
REQUIRE(a3.WeakLearner(0).MaxIterations() == 150);
REQUIRE(a4.WeakLearner(0).MaxIterations() == 1000);
REQUIRE(a5.WeakLearner(0).MaxIterations() == 1000);
REQUIRE(a6.WeakLearner(0).MaxIterations() == 1000);
REQUIRE(a7.WeakLearner(0).MaxIterations() == 100);
}