Adding changes in implementation

mlpack · May 14, 2020 · c6a8f87 · c6a8f87
1 parent 0a8e983
commit c6a8f87
Show file tree

Hide file tree

Showing 3 changed files with 60 additions and 53 deletions.
diff --git a/src/mlpack/methods/ann/layer/layer_types.hpp b/src/mlpack/methods/ann/layer/layer_types.hpp
@@ -252,7 +252,6 @@ using LayerTypes = boost::variant<
     LeakyReLU<arma::mat, arma::mat>*,
     CReLU<arma::mat, arma::mat>*,
     Linear<arma::mat, arma::mat, NoRegularizer>*,
-    RBF<arma::mat, arma::mat>*,
     LinearNoBias<arma::mat, arma::mat, NoRegularizer>*,
     LogSoftMax<arma::mat, arma::mat>*,
     Lookup<arma::mat, arma::mat>*,

diff --git a/src/mlpack/methods/ann/layer/radial_basis_function_impl.hpp b/src/mlpack/methods/ann/layer/radial_basis_function_impl.hpp
@@ -41,19 +41,27 @@ void RBF<InputDataType, OutputDataType>::Forward(
     const arma::Mat<eT>& input,
     arma::Mat<eT>& output)
 {
-  centres = arma::mat(outSize, input.n_rows, arma::fill::randu);
+  centres = arma::mat(input.n_rows, outSize, arma::fill::randu);
   centres = arma::normcdf(centres, 0, 1);
-  sigmas = arma::ones(1, outSize);
+  sigmas = arma::ones(outSize, 1);
+  arma::cube x = arma::cube(input.n_rows, outSize, input.n_cols);
 
-  distances = arma::mat(outSize, input.n_cols);
+  for (size_t i = 0; i < input.n_cols; i++)
+  {
+    x.slice(i).each_col() = input.col(i);
+  }
+
+  arma::cube c = arma::cube(input.n_rows, outSize, input.n_cols);
+  c.each_slice() = centres;
+
+  distances = arma::mat(input.n_rows, input.n_cols);
 
-  for (size_t i = 0; i < outSize; i++)
+  for (size_t i = 0; i < input.n_cols; i++)
   {
-    arma::mat temp = centres.each_col(i)
     distances.col(i) = arma::pow(arma::sum(
                                  arma::pow((
-                                 temp),
-                                 2), 1), 0.5).t() * sigmas(i);
+                                 x.slice(i) - c.slice(i)),
+                                 2), 1), 0.5);
   }
 
   output = distances;

diff --git a/src/mlpack/tests/feedforward_network_test.cpp b/src/mlpack/tests/feedforward_network_test.cpp
@@ -624,20 +624,20 @@ BOOST_AUTO_TEST_CASE(OptimizerTest)
 /**
  * Train the RBF network on a larger dataset.
  */
-//BOOST_AUTO_TEST_CASE(RBFNetworkTest)
-//{
-//  // Load the dataset.
-//  arma::mat trainData;
-//  data::Load("thyroid_train.csv", trainData, true);
-//
-//  arma::mat trainLabels = trainData.row(trainData.n_rows - 1);
-//  trainData.shed_row(trainData.n_rows - 1);
-//
-//  arma::mat testData;
-//  data::Load("thyroid_test.csv", testData, true);
-//
-// arma::mat testLabels = testData.row(testData.n_rows - 1);
-//  testData.shed_row(testData.n_rows - 1);
+BOOST_AUTO_TEST_CASE(RBFNetworkTest)
+{
+  // Load the dataset.
+  arma::mat trainData;
+  data::Load("thyroid_train.csv", trainData, true);
+
+  arma::mat trainLabels = trainData.row(trainData.n_rows - 1);
+  trainData.shed_row(trainData.n_rows - 1);
+
+  arma::mat testData;
+  data::Load("thyroid_test.csv", testData, true);
+
+  arma::mat testLabels = testData.row(testData.n_rows - 1);
+  testData.shed_row(testData.n_rows - 1);
 
   /*
    * Construct a feed forward network with trainData.n_rows input nodes,
@@ -652,36 +652,36 @@ BOOST_AUTO_TEST_CASE(OptimizerTest)
    * |     |       |     |       |     |       |     |
    * +-----+       +--+--+       +-----+       +-----+
    */
-//
-//  FFN<NegativeLogLikelihood<> > model;
-//  model.Add<RBF<> >(trainData.n_cols, 8);
-//  model.Add<GaussianFunctionLayer<> >();
-//  model.Add<Linear<> >(trainData.n_rows, 8);
-//  model.Add<Linear<> >(8, 3);
-//  model.Add<LogSoftMax<> >();
-//
-//  TestNetwork<>(model, trainData, trainLabels, testData, testLabels, 10, 0.1);
-//  arma::mat dataset;
-//  dataset.load("mnist_first250_training_4s_and_9s.arm");
-//
-//  // Normalize each point since these are images.
-//  for (size_t i = 0; i < dataset.n_cols; ++i)
-//  {
-//    dataset.col(i) /= norm(dataset.col(i), 2);
-//  }
-//
-//  arma::mat labels = arma::zeros(1, dataset.n_cols);
-//  labels.submat(0, labels.n_cols / 2, 0, labels.n_cols - 1).fill(1);
-//  labels += 1;
-//
-//  FFN<NegativeLogLikelihood<> > model1;
-//  model1.Add<RBF<> >(dataset.n_cols, 10);
-//  model1.Add<GaussianFunctionLayer<> >();
-//  model1.Add<Linear<> >(dataset.n_rows, 10);
-//  model.Add<Linear<> >(10, 2);
-//  model1.Add<LogSoftMax<> >();
-//  // Vanilla neural net with logistic activation function.
-//  TestNetwork<>(model1, dataset, labels, dataset, labels, 10, 0.2);
-//}
+
+  FFN<NegativeLogLikelihood<> > model;
+  model.Add<RBF<> >(trainData.n_cols, 8);
+  model.Add<GaussianFunctionLayer<> >();
+  model.Add<Linear<> >(trainData.n_rows, 8);
+  model.Add<Linear<> >(8, 3);
+  model.Add<LogSoftMax<> >();
+
+  TestNetwork<>(model, trainData, trainLabels, testData, testLabels, 10, 0.1);
+  arma::mat dataset;
+  dataset.load("mnist_first250_training_4s_and_9s.arm");
+
+  // Normalize each point since these are images.
+  for (size_t i = 0; i < dataset.n_cols; ++i)
+  {
+    dataset.col(i) /= norm(dataset.col(i), 2);
+  }
+
+  arma::mat labels = arma::zeros(1, dataset.n_cols);
+  labels.submat(0, labels.n_cols / 2, 0, labels.n_cols - 1).fill(1);
+  labels += 1;
+
+  FFN<NegativeLogLikelihood<> > model1;
+  model1.Add<RBF<> >(dataset.n_cols, 10);
+  model1.Add<GaussianFunctionLayer<> >();
+  model1.Add<Linear<> >(dataset.n_rows, 10);
+  model.Add<Linear<> >(10, 2);
+  model1.Add<LogSoftMax<> >();
+  // Vanilla neural net with logistic activation function.
+  TestNetwork<>(model1, dataset, labels, dataset, labels, 10, 0.2);
+}
 
 BOOST_AUTO_TEST_SUITE_END();