1 : add read only function FeatureSize

2 : add function InitializeWeights to avoid the case when RVO fail

src/mlpack/methods/softmax_regression/softmax_regression.hpp

@@ -161,6 +161,10 @@ class SoftmaxRegression
   //! Get the model parameters.
   const arma::mat& Parameters() const { return parameters; }
 
+  //! Gets the features size of the training data
+  size_t FeatureSize() const
+  { return parameters.n_rows; }
+
   /**
    * Serialize the SoftmaxRegression model. 
    */

src/mlpack/methods/softmax_regression/softmax_regression_function.cpp

@@ -40,20 +40,28 @@ const arma::mat SoftmaxRegressionFunction::
 InitializeWeights(const size_t featureSize,
                   const size_t numClasses,
                   const bool fitIntercept)
-{
-    // Initialize values to 0.005 * r. 'r' is a matrix of random values taken from
-    // a Gaussian distribution with mean zero and variance one.
-    // If the fitIntercept flag is true, parameters.col(0) is the intercept.
+{    
     arma::mat parameters;
-    if (fitIntercept)
-      parameters.randn(numClasses, featureSize + 1);
-    else
-      parameters.randn(numClasses, featureSize);
-    parameters = 0.005 * parameters;
-
+    InitializeWeights(parameters, featureSize, numClasses, fitIntercept);
     return parameters;
 }
 
+void SoftmaxRegressionFunction::
+InitializeWeights(arma::mat &weights,
+                  const size_t featureSize,
+                  const size_t numClasses,
+                  const bool fitIntercept)
+{
+  // Initialize values to 0.005 * r. 'r' is a matrix of random values taken from
+  // a Gaussian distribution with mean zero and variance one.
+  // If the fitIntercept flag is true, parameters.col(0) is the intercept.
+  if (fitIntercept)
+    weights.randn(numClasses, featureSize + 1);
+  else
+    weights.randn(numClasses, featureSize);
+  weights *= 0.005;
+}
+
 /**
  * This is equivalent to applying the indicator function to the training
  * labels. The output is in the form of a matrix, which leads to simpler

src/mlpack/methods/softmax_regression/softmax_regression_function.hpp

@@ -48,6 +48,20 @@ class SoftmaxRegressionFunction
                                            const size_t numClasses,
                                            const bool fitIntercept = false);
 
+  /**
+   * Initialize Softmax Regression weights(trainable parameters) with
+   * the given parameters.
+   * @paaram weights weights want to initialize
+   * @param featureSize The features size of the training set
+   * @param numClasses Number of classes for classification.
+   * @param fitIntercept Intercept term flag.
+   * @return weights after initialize
+   */
+  static void InitializeWeights(arma::mat &weights,
+                                const size_t featureSize,
+                                const size_t numClasses,
+                                const bool fitIntercept = false);
+
   /**
    * Constructs the ground truth label matrix with the passed labels.
    *
@@ -98,7 +112,8 @@ class SoftmaxRegressionFunction
   size_t NumClasses() const { return numClasses; }
 
   //! Gets the features size of the training data
-  size_t FeatureSize() const { return data.n_rows; }
+  size_t FeatureSize() const
+  { return initialPoint.n_rows; }
 
   //! Sets the regularization parameter.
   double& Lambda() { return lambda; }

src/mlpack/methods/softmax_regression/softmax_regression_impl.hpp

@@ -22,8 +22,9 @@ SoftmaxRegression(const size_t inputSize,
     lambda(0.0001),
     fitIntercept(fitIntercept)
 {  
-  parameters = SoftmaxRegressionFunction::InitializeWeights(inputSize, numClasses,
-                                                            fitIntercept);
+   SoftmaxRegressionFunction::InitializeWeights(parameters,
+                                                inputSize, numClasses,
+                                                fitIntercept);
 }
 
 template<template<typename> class OptimizerType>