FeatureTransformEstimateMulti Class Reference

#include <get-feature-transform.h>

Inheritance diagram for FeatureTransformEstimateMulti:

Collaboration diagram for FeatureTransformEstimateMulti:

Public Member Functions
void	Estimate (const FeatureTransformEstimateOptions &opts, const std::vector< std::vector< int32 > > &indexes, Matrix< BaseFloat > *M) const
	This is as FeatureTransformEstimate, but for use in nnet-get-feature-transform-multi.cc, see the usage message of that program for a description of what it does. More...
Public Member Functions inherited from FeatureTransformEstimate
void	Estimate (const FeatureTransformEstimateOptions &opts, Matrix< BaseFloat > *M, TpMatrix< BaseFloat > *within_cholesky) const
	Estimates the LDA transform matrix m. More...
Public Member Functions inherited from LdaEstimate
	LdaEstimate ()
void	Init (int32 num_classes, int32 dimension)
	Allocates memory for accumulators. More...
int32	NumClasses () const
	Returns the number of classes. More...
int32	Dim () const
	Returns the dimensionality of the feature vectors. More...
void	ZeroAccumulators ()
	Sets all accumulators to zero. More...
void	Scale (BaseFloat f)
	Scales all accumulators. More...
double	TotCount ()
	Return total count of the data. More...
void	Accumulate (const VectorBase< BaseFloat > &data, int32 class_id, BaseFloat weight=1.0)
	Accumulates data. More...
void	Estimate (const LdaEstimateOptions &opts, Matrix< BaseFloat > *M, Matrix< BaseFloat > *Mfull=NULL) const
	Estimates the LDA transform matrix m. More...
void	Read (std::istream &in_stream, bool binary, bool add)
void	Write (std::ostream &out_stream, bool binary) const

Private Member Functions
void	EstimateTransformPart (const FeatureTransformEstimateOptions &opts, const std::vector< int32 > &indexes, const SpMatrix< double > &total_covar, const SpMatrix< double > &between_covar, const Vector< double > &mean, Matrix< BaseFloat > *M) const

Additional Inherited Members
Protected Member Functions inherited from LdaEstimate
void	GetStats (SpMatrix< double > *total_covar, SpMatrix< double > *between_covar, Vector< double > *total_mean, double *sum) const
	Extract a more processed form of the stats. More...
LdaEstimate &	operator= (const LdaEstimate &other)
Static Protected Member Functions inherited from FeatureTransformEstimate
static void	EstimateInternal (const FeatureTransformEstimateOptions &opts, const SpMatrix< double > &total_covar, const SpMatrix< double > &between_covar, const Vector< double > &mean, Matrix< BaseFloat > *M, TpMatrix< BaseFloat > *C)
Static Protected Member Functions inherited from LdaEstimate
static void	AddMeanOffset (const VectorBase< double > &total_mean, Matrix< BaseFloat > *projection)
	This function modifies the LDA matrix so that it also subtracts the mean feature value. More...
Protected Attributes inherited from LdaEstimate
Vector< double >	zero_acc_
Matrix< double >	first_acc_
SpMatrix< double >	total_second_acc_

Detailed Description

Definition at line 157 of file get-feature-transform.h.

Member Function Documentation

Estimate()

void Estimate	(	const FeatureTransformEstimateOptions &	opts,
		const std::vector< std::vector< int32 > > &	indexes,
		Matrix< BaseFloat > *	M
	)		const

This is as FeatureTransformEstimate, but for use in nnet-get-feature-transform-multi.cc, see the usage message of that program for a description of what it does.

Definition at line 165 of file get-feature-transform.cc.

References count, LdaEstimate::Dim(), LdaEstimate::GetStats(), rnnlm::i, kaldi::IsSortedAndUniq(), KALDI_ASSERT, MatrixBase< Real >::NumCols(), MatrixBase< Real >::Range(), FeatureTransformEstimateOptions::remove_offset, and Matrix< Real >::Resize().

Referenced by main().

                                {

  int32 input_dim = Dim(), output_dim = 0, num_transforms = indexes.size();
  for (int32 i = 0; i < num_transforms; i++) { // some input-checking.
    KALDI_ASSERT(indexes[i].size() > 0);
    std::vector<int32> this_indexes(indexes[i]);
    std::sort(this_indexes.begin(), this_indexes.end());
    KALDI_ASSERT(IsSortedAndUniq(this_indexes)); // check for duplicates.
    KALDI_ASSERT(this_indexes.front() >= 0);
    KALDI_ASSERT(this_indexes.back() < input_dim);
    output_dim += this_indexes.size();
  }

  int32 input_dim_ext = (opts.remove_offset ? input_dim + 1 : input_dim);
  M->Resize(output_dim, input_dim_ext);
  
  double count;
  Vector<double> total_mean;
  SpMatrix<double> total_covar, between_covar;
  GetStats(&total_covar, &between_covar, &total_mean, &count);

  int32 cur_output_index = 0;
  for (int32 i = 0; i < num_transforms; i++) {
    Matrix<BaseFloat> M_tmp;
    EstimateTransformPart(opts, indexes[i], total_covar, between_covar,
                          total_mean, &M_tmp);
    int32 this_output_dim = indexes[i].size();
    M->Range(cur_output_index, this_output_dim, 0, M->NumCols()).
        CopyFromMat(M_tmp);
    cur_output_index += this_output_dim;
  }
  
}

EstimateTransformPart()

void EstimateTransformPart ( const FeatureTransformEstimateOptions &  opts, const std::vector< int32 > &  indexes, const SpMatrix< double > &  total_covar, const SpMatrix< double > &  between_covar, const Vector< double > &  mean, Matrix< BaseFloat > *  M  ) const

private

Definition at line 130 of file get-feature-transform.cc.

References SpMatrix< Real >::AddMat2Sp(), MatrixBase< Real >::AddMatMat(), VectorBase< Real >::AddMatVec(), FeatureTransformEstimateOptions::dim, LdaEstimate::Dim(), FeatureTransformEstimate::EstimateInternal(), rnnlm::i, kaldi::kCopyData, kaldi::kNoTrans, MatrixBase< Real >::NumCols(), and Matrix< Real >::Resize().

                                {

  int32 full_dim = Dim(), proj_dim = indexes.size();
  Matrix<double> transform(proj_dim, full_dim); // projects from full to projected dim.
  for (int32 i = 0; i < proj_dim; i++)
    transform(i, indexes[i]) = 1.0;

  SpMatrix<double> total_covar_proj(proj_dim), between_covar_proj(proj_dim);
  Vector<double> mean_proj(proj_dim);
  total_covar_proj.AddMat2Sp(1.0, transform, kNoTrans, total_covar, 0.0);
  between_covar_proj.AddMat2Sp(1.0, transform, kNoTrans, between_covar, 0.0);
  mean_proj.AddMatVec(1.0, transform, kNoTrans, mean, 0.0);

  Matrix<BaseFloat> M_proj;
  FeatureTransformEstimateOptions opts_tmp(opts);
  opts_tmp.dim = proj_dim;
  EstimateInternal(opts_tmp, total_covar_proj, between_covar_proj, mean_proj,
                   &M_proj, NULL);
  if (M_proj.NumCols() == proj_dim + 1) { // Extend transform to add the extra "1" that we
                                          // use to handle mean shifts..
    transform.Resize(proj_dim + 1, full_dim + 1, kCopyData);
    transform(proj_dim, full_dim) = 1.0;
  }
  M->Resize(proj_dim, transform.NumCols());
  // Produce output..
  M->AddMatMat(1.0, M_proj, kNoTrans, Matrix<BaseFloat>(transform),
               kNoTrans, 0.0);
}

---

The documentation for this class was generated from the following files:

• nnet2/get-feature-transform.h
• nnet2/get-feature-transform.cc