Added extra argument to a couple of matrix-library functions.

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@1363 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8
This commit is contained in:
Dan Povey 2012-09-18 05:09:15 +00:00
Родитель 45e4d2f0a0
Коммит 8163148b37
20 изменённых файлов: 232 добавлений и 171 удалений

Просмотреть файл

@ -310,9 +310,9 @@ void CuVector<Real>::AddRowSumMat(Real alpha, const CuMatrix<Real> &mat, Real be
#endif #endif
{ {
Vector<Real> tmp(mat.NumCols()); Vector<Real> tmp(mat.NumCols());
tmp.AddRowSumMat(mat.Mat()); tmp.AddRowSumMat(1.0, mat.Mat());
if(beta != 1.0) vec_.Scale(beta); if(beta != 1.0) vec_.Scale(beta);
vec_.AddVec(alpha,tmp); vec_.AddVec(alpha, tmp);
} }
} }
@ -360,7 +360,7 @@ void CuVector<Real>::AddColSumMat(Real alpha, const CuMatrix<Real> &mat, Real be
#endif #endif
{ {
Vector<Real> tmp(mat.NumRows()); Vector<Real> tmp(mat.NumRows());
tmp.AddColSumMat(mat.Mat()); tmp.AddColSumMat(1.0, mat.Mat());
if(beta != 1.0) vec_.Scale(beta); if(beta != 1.0) vec_.Scale(beta);
vec_.AddVec(alpha,tmp); vec_.AddVec(alpha,tmp);
} }

Просмотреть файл

@ -153,7 +153,7 @@ int main(int argc, char *argv[]) {
} }
if (subtract_mean) { if (subtract_mean) {
Vector<BaseFloat> mean(features.NumCols()); Vector<BaseFloat> mean(features.NumCols());
mean.AddRowSumMat(features); mean.AddRowSumMat(1.0, features);
mean.Scale(1.0 / features.NumRows()); mean.Scale(1.0 / features.NumRows());
for (int32 i = 0; i < features.NumRows(); i++) for (int32 i = 0; i < features.NumRows(); i++)
features.Row(i).AddVec(-1.0, mean); features.Row(i).AddVec(-1.0, mean);

Просмотреть файл

@ -153,7 +153,7 @@ int main(int argc, char *argv[]) {
} }
if (subtract_mean) { if (subtract_mean) {
Vector<BaseFloat> mean(features.NumCols()); Vector<BaseFloat> mean(features.NumCols());
mean.AddRowSumMat(features); mean.AddRowSumMat(1.0, features);
mean.Scale(1.0 / features.NumRows()); mean.Scale(1.0 / features.NumRows());
for (int32 i = 0; i < features.NumRows(); i++) for (int32 i = 0; i < features.NumRows(); i++)
features.Row(i).AddVec(-1.0, mean); features.Row(i).AddVec(-1.0, mean);

Просмотреть файл

@ -153,7 +153,7 @@ int main(int argc, char *argv[]) {
} }
if (subtract_mean) { if (subtract_mean) {
Vector<BaseFloat> mean(features.NumCols()); Vector<BaseFloat> mean(features.NumCols());
mean.AddRowSumMat(features); mean.AddRowSumMat(1.0, features);
mean.Scale(1.0 / features.NumRows()); mean.Scale(1.0 / features.NumRows());
for (size_t i = 0; i < features.NumRows(); i++) for (size_t i = 0; i < features.NumRows(); i++)
features.Row(i).AddVec(-1.0, mean); features.Row(i).AddVec(-1.0, mean);

Просмотреть файл

@ -116,7 +116,7 @@ int main(int argc, char *argv[]) {
} }
if (subtract_mean) { if (subtract_mean) {
Vector<BaseFloat> mean(features.NumCols()); Vector<BaseFloat> mean(features.NumCols());
mean.AddRowSumMat(features); mean.AddRowSumMat(1.0, features);
mean.Scale(1.0 / features.NumRows()); mean.Scale(1.0 / features.NumRows());
for (int32 i = 0; i < features.NumRows(); i++) for (int32 i = 0; i < features.NumRows(); i++)
features.Row(i).AddVec(-1.0, mean); features.Row(i).AddVec(-1.0, mean);

Просмотреть файл

@ -52,7 +52,7 @@ int main(int argc, char *argv[]) {
continue; continue;
} }
Vector<BaseFloat> mean(feats.NumCols()); Vector<BaseFloat> mean(feats.NumCols());
mean.AddRowSumMat(feats); mean.AddRowSumMat(1.0, feats);
mean.Scale(1.0 / feats.NumRows()); mean.Scale(1.0 / feats.NumRows());
for (int32 i = 0; i < feats.NumRows(); i++) for (int32 i = 0; i < feats.NumRows(); i++)
feats.Row(i).AddVec(-1.0, mean); feats.Row(i).AddVec(-1.0, mean);

Просмотреть файл

@ -309,7 +309,7 @@ UnitTestEstimateFullGmm() {
Vector<BaseFloat> mean(dim); Vector<BaseFloat> mean(dim);
cov.AddMatMat(1.0, feats, kTrans, feats, kNoTrans, 0.0); cov.AddMatMat(1.0, feats, kTrans, feats, kNoTrans, 0.0);
cov.Scale(1.0 / feats.NumRows()); cov.Scale(1.0 / feats.NumRows());
mean.AddRowSumMat(feats); mean.AddRowSumMat(1.0, feats);
mean.Scale(1.0 / feats.NumRows()); mean.Scale(1.0 / feats.NumRows());
cov.AddVecVec(-1.0, mean, mean); cov.AddVecVec(-1.0, mean, mean);
BaseFloat logdet = cov.LogDet(); BaseFloat logdet = cov.LogDet();

Просмотреть файл

@ -763,38 +763,55 @@ void MatrixBase<double>::CopyFromTp(const TpMatrix<double> & M,
template<typename Real> template<typename Real>
void MatrixBase<Real>::CopyRowsFromVec(const VectorBase<Real> &rv) { void MatrixBase<Real>::CopyRowsFromVec(const VectorBase<Real> &rv) {
KALDI_ASSERT(rv.Dim() == num_rows_*num_cols_); if (rv.Dim() == num_rows_*num_cols_) {
if (stride_ == num_cols_) {
if (stride_ == num_cols_) { // one big copy operation.
// one big copy operation. const Real *rv_data = rv.Data();
const Real *rv_data = rv.Data(); std::memcpy(data_, rv_data, sizeof(Real)*num_rows_*num_cols_);
std::memcpy(data_, rv_data, sizeof(Real)*num_rows_*num_cols_); } else {
} else { const Real *rv_data = rv.Data();
const Real *rv_data = rv.Data(); for (MatrixIndexT r = 0; r < num_rows_; r++) {
for (MatrixIndexT r = 0; r < num_rows_; r++) { Real *row_data = RowData(r);
Real *row_data = RowData(r); for (MatrixIndexT c = 0; c < num_cols_; c++) {
for (MatrixIndexT c = 0; c < num_cols_; c++) { row_data[c] = rv_data[c];
row_data[c] = rv_data[c]; }
rv_data += num_cols_;
} }
rv_data += num_cols_;
} }
} else if (rv.Dim() == num_cols_) {
const Real *rv_data = rv.Data();
for (MatrixIndexT r = 0; r < num_rows_; r++)
std::memcpy(RowData(r), rv_data, sizeof(Real)*num_cols_);
} else {
KALDI_ERR << "Wrong sized arguments";
} }
} }
template<typename Real> template<typename Real>
template<typename OtherReal> template<typename OtherReal>
void MatrixBase<Real>::CopyRowsFromVec(const VectorBase<OtherReal> &rv) { void MatrixBase<Real>::CopyRowsFromVec(const VectorBase<OtherReal> &rv) {
KALDI_ASSERT(rv.Dim() == num_rows_*num_cols_); if (rv.Dim() == num_rows_*num_cols_) {
const OtherReal *rv_data = rv.Data(); const OtherReal *rv_data = rv.Data();
for (MatrixIndexT r = 0; r < num_rows_; r++) { for (MatrixIndexT r = 0; r < num_rows_; r++) {
Real *row_data = RowData(r); Real *row_data = RowData(r);
for (MatrixIndexT c = 0; c < num_cols_; c++) { for (MatrixIndexT c = 0; c < num_cols_; c++) {
row_data[c] = static_cast<Real>(rv_data[c]); row_data[c] = static_cast<Real>(rv_data[c]);
}
rv_data += num_cols_;
} }
rv_data += num_cols_; } else if (rv.Dim() == num_cols_) {
const OtherReal *rv_data = rv.Data();
Real *first_row_data = RowData(0);
for (MatrixIndexT c = 0; c < num_cols_; c++)
first_row_data[c] = rv_data[c];
for (MatrixIndexT r = 1; r < num_rows_; r++)
std::memcpy(RowData(r), first_row_data, sizeof(Real)*num_cols_);
} else {
KALDI_ERR << "Wrong sized arguments.";
} }
} }
template template
void MatrixBase<float>::CopyRowsFromVec(const VectorBase<double> &rv); void MatrixBase<float>::CopyRowsFromVec(const VectorBase<double> &rv);
template template
@ -802,17 +819,28 @@ void MatrixBase<double>::CopyRowsFromVec(const VectorBase<float> &rv);
template<typename Real> template<typename Real>
void MatrixBase<Real>::CopyColsFromVec(const VectorBase<Real> &rv) { void MatrixBase<Real>::CopyColsFromVec(const VectorBase<Real> &rv) {
KALDI_ASSERT(rv.Dim() == num_rows_*num_cols_); if (rv.Dim() == num_rows_*num_cols_) {
const Real *v_inc_data = rv.Data();
Real *m_inc_data = data_;
const Real *v_inc_data = rv.Data(); for (MatrixIndexT c = 0; c < num_cols_; c++) {
Real *m_inc_data = data_; for (MatrixIndexT r = 0; r < num_rows_; r++) {
m_inc_data[r * stride_] = v_inc_data[r];
for (MatrixIndexT c = 0; c < num_cols_; c++) { }
for (MatrixIndexT r = 0; r < num_rows_; r++) { v_inc_data += num_rows_;
m_inc_data[r * stride_] = v_inc_data[r]; m_inc_data ++;
} }
v_inc_data += num_rows_; } else if (rv.Dim() == num_rows_) {
m_inc_data ++; const Real *v_inc_data = rv.Data();
Real *m_inc_data = data_;
for (MatrixIndexT r = 0; r < num_rows_; r++) {
BaseFloat value = *(v_inc_data++);
for (MatrixIndexT c = 0; c < num_cols_; c++)
m_inc_data[c] = value;
m_inc_data += stride_;
}
} else {
KALDI_ERR << "Wrong size of arguments.";
} }
} }

Просмотреть файл

@ -130,14 +130,17 @@ class MatrixBase {
MatrixTransposeType Trans = kNoTrans); MatrixTransposeType Trans = kNoTrans);
/// Inverse of vec() operator. Copies vector into matrix, row-by-row. /// Inverse of vec() operator. Copies vector into matrix, row-by-row.
/// Note that rv.Dim() must equal NumRows()*NumCols(). /// Note that rv.Dim() must either equal NumRows()*NumCols() or
/// NumCols()-- this has two modes of operation.
void CopyRowsFromVec(const VectorBase<Real> &v); void CopyRowsFromVec(const VectorBase<Real> &v);
template<typename OtherReal> template<typename OtherReal>
void CopyRowsFromVec(const VectorBase<OtherReal> &v); void CopyRowsFromVec(const VectorBase<OtherReal> &v);
/// Copies vector into matrix, column-by-column. /// Copies vector into matrix, column-by-column.
/// Note that rv.Dim() must equal NumRows()*NumCols(). /// Note that rv.Dim() must either equal NumRows()*NumCols() or NumRows();
/// this has two modes of operation.
void CopyColsFromVec(const VectorBase<Real> &v); void CopyColsFromVec(const VectorBase<Real> &v);
/// Copy vector into specific column of matrix. /// Copy vector into specific column of matrix.
void CopyColFromVec(const VectorBase<Real> &v, const MatrixIndexT col); void CopyColFromVec(const VectorBase<Real> &v, const MatrixIndexT col);
/// Copy vector into specific row of matrix. /// Copy vector into specific row of matrix.
@ -745,6 +748,12 @@ std::istream & operator >> (std::istream & In, MatrixBase<Real> & M);
template<typename Real> template<typename Real>
std::istream & operator >> (std::istream & In, Matrix<Real> & M); std::istream & operator >> (std::istream & In, Matrix<Real> & M);
template<class Real>
bool SameDim(const MatrixBase<Real> &M, const MatrixBase<Real> &N) {
return (M.NumRows() == N.NumRows() && M.NumCols() == N.NumCols());
}
/// @} end of \addtogroup matrix_funcs_io /// @} end of \addtogroup matrix_funcs_io

Просмотреть файл

@ -551,7 +551,7 @@ Real VectorBase<Real>::SumLog() const {
} }
template<typename Real> template<typename Real>
void VectorBase<Real>::AddRowSumMat(const MatrixBase<Real> &rM) { void VectorBase<Real>::AddRowSumMat(Real alpha, const MatrixBase<Real> &rM) {
// note the double accumulator // note the double accumulator
double sum; double sum;
KALDI_ASSERT(dim_ == rM.NumCols()); KALDI_ASSERT(dim_ == rM.NumCols());
@ -560,12 +560,12 @@ void VectorBase<Real>::AddRowSumMat(const MatrixBase<Real> &rM) {
for (MatrixIndexT j = 0; j < rM.NumRows(); j++) { for (MatrixIndexT j = 0; j < rM.NumRows(); j++) {
sum += rM(j, i); sum += rM(j, i);
} }
data_[i] += sum; data_[i] += alpha * sum;
} }
} }
template<typename Real> template<typename Real>
void VectorBase<Real>::AddColSumMat(const MatrixBase<Real> &rM) { void VectorBase<Real>::AddColSumMat(Real alpha, const MatrixBase<Real> &rM) {
// note the double accumulator // note the double accumulator
double sum; double sum;
KALDI_ASSERT(dim_ == rM.NumRows()); KALDI_ASSERT(dim_ == rM.NumRows());
@ -574,7 +574,7 @@ void VectorBase<Real>::AddColSumMat(const MatrixBase<Real> &rM) {
for (MatrixIndexT j = 0; j < rM.NumCols(); j++) { for (MatrixIndexT j = 0; j < rM.NumCols(); j++) {
sum += rM(i, j); sum += rM(i, j);
} }
data_[i] += sum; data_[i] += alpha * sum;
} }
} }

Просмотреть файл

@ -242,11 +242,11 @@ class VectorBase {
/// negative. /// negative.
Real SumLog() const; Real SumLog() const;
/// Adds sum of the rows of M to existing contents. /// Adds sum of the rows of M to existing contents, times alpha.
void AddRowSumMat(const MatrixBase<Real>& M); void AddRowSumMat(Real alpha, const MatrixBase<Real>& M);
/// Adds sum of the columns of M to existing contents. /// Adds sum of the columns of M to existing contents.
void AddColSumMat(const MatrixBase<Real>& M); void AddColSumMat(Real alpha, const MatrixBase<Real>& M);
/// Returns log(sum(exp())) without exp overflow /// Returns log(sum(exp())) without exp overflow
/// If prune > 0.0, ignores terms less than the max - prune. /// If prune > 0.0, ignores terms less than the max - prune.

Просмотреть файл

@ -67,16 +67,16 @@ start:
for (MatrixIndexT i = 0;i < M->NumRows();i++) for (MatrixIndexT i = 0;i < M->NumRows();i++)
for (MatrixIndexT j = 0;j < M->NumCols();j++) for (MatrixIndexT j = 0;j < M->NumCols();j++)
(*M)(i, j) = RandGauss(); (*M)(i, j) = RandGauss();
if (M->NumRows() != 0 && M->Cond() > 100) { if (M->NumRows() != 0 && M->Cond() > 100) {
printf("Condition number of random matrix large %f, trying again (this is normal)\n", printf("Condition number of random matrix large %f, trying again (this is normal)\n",
(float) M->Cond()); (float) M->Cond());
goto start; goto start;
} }
} }
template<class Real> static void InitRand(SpMatrix<Real> *M) { template<class Real> static void InitRand(SpMatrix<Real> *M) {
start: start:
for (MatrixIndexT i = 0;i < M->NumRows();i++) for (MatrixIndexT i = 0;i < M->NumRows();i++)
for (MatrixIndexT j = 0;j<=i;j++) for (MatrixIndexT j = 0;j<=i;j++)
(*M)(i, j) = RandGauss(); (*M)(i, j) = RandGauss();
@ -115,14 +115,14 @@ static bool ApproxEqual(const SpMatrix<Real> &A,
} }
/* was: /* was:
template<class Real> template<class Real>
bool ApproxEqual(SpMatrix<Real> &A, SpMatrix<Real> &B, float tol = 0.001) { bool ApproxEqual(SpMatrix<Real> &A, SpMatrix<Real> &B, float tol = 0.001) {
KALDI_ASSERT(A.NumRows() == B.NumRows()&&A.NumCols() == B.NumCols()); KALDI_ASSERT(A.NumRows() == B.NumRows()&&A.NumCols() == B.NumCols());
for (MatrixIndexT i = 0;i < A.NumRows();i++) for (MatrixIndexT i = 0;i < A.NumRows();i++)
for (MatrixIndexT j = 0;j<=i;j++) for (MatrixIndexT j = 0;j<=i;j++)
if (std::abs(A(i, j)-B(i, j)) > tol*std::max(1.0, (double) (std::abs(A(i, j))+std::abs(B(i, j))))) return false; if (std::abs(A(i, j)-B(i, j)) > tol*std::max(1.0, (double) (std::abs(A(i, j))+std::abs(B(i, j))))) return false;
return true; return true;
} }
*/ */
template<class Real> static void AssertEqual(Vector<Real> &A, Vector<Real> &B, float tol = 0.001) { template<class Real> static void AssertEqual(Vector<Real> &A, Vector<Real> &B, float tol = 0.001) {
@ -209,6 +209,27 @@ static void UnitTestSpAddVec() {
} }
} }
template<class Real> static void UnitTestCopyRowsAndCols() {
// Test other mode of CopyRowsFromVec, and CopyColsFromVec,
// where vector is duplicated.
for (int32 i = 0; i < 30; i++) {
int32 dimM = 1 + rand() % 5, dimN = 1 + rand() % 5;
Vector<float> w(dimN); // test cross-type version of
// CopyRowsFromVec.
Vector<Real> v(dimM);
Matrix<Real> M(dimM, dimN), N(dimM, dimN);
InitRand(&v);
InitRand(&w);
M.CopyColsFromVec(v);
N.CopyRowsFromVec(w);
for (int32 r = 0; r < dimM; r++) {
for (int32 c = 0; c < dimN; c++) {
KALDI_ASSERT(M(r, c) == v(r));
KALDI_ASSERT(N(r, c) == w(c));
}
}
}
}
template<class Real> static void UnitTestSpliceRows() { template<class Real> static void UnitTestSpliceRows() {
@ -291,10 +312,10 @@ template<class Real> static void UnitTestRemoveRow() {
Vector<Real> N(V); Vector<Real> N(V);
N.RemoveElement(i); N.RemoveElement(i);
for (MatrixIndexT j = 0;j < i;j++) { for (MatrixIndexT j = 0;j < i;j++) {
KALDI_ASSERT(V(j) == N(j)); KALDI_ASSERT(V(j) == N(j));
} }
for (MatrixIndexT j = i+1;j < dimM;j++) { for (MatrixIndexT j = i+1;j < dimM;j++) {
KALDI_ASSERT(V(j) == N(j-1)); KALDI_ASSERT(V(j) == N(j-1));
} }
} }
@ -354,8 +375,10 @@ static void UnitTestSimpleForVec() { // testing some simple operaters on vector
Matrix<Real> M(dimM, dimN); Matrix<Real> M(dimM, dimN);
InitRand(&M); InitRand(&M);
Vector<Real> Vr(dimN), Vc(dimM); Vector<Real> Vr(dimN), Vc(dimM);
Vr.AddRowSumMat(M); Vr.AddRowSumMat(0.5, M);
Vc.AddColSumMat(M); Vc.AddColSumMat(0.5, M);
Vr.Scale(2.0);
Vc.Scale(2.0);
Vector<Real> V2r(dimN), V2c(dimM); Vector<Real> V2r(dimN), V2c(dimM);
for (MatrixIndexT k = 0; k < dimM; k++) { for (MatrixIndexT k = 0; k < dimM; k++) {
@ -511,7 +534,7 @@ static void UnitTestSimpleForMat() { // test some simple operates on all kinds
y.Cholesky(x); y.Cholesky(x);
std::cout << "Matrix y is a lower triangular Cholesky decomposition of x:" std::cout << "Matrix y is a lower triangular Cholesky decomposition of x:"
<< '\n'; << '\n';
std::cout << y << '\n'; std::cout << y << '\n';
// test sp-matrix's LogPosDefDet() function // test sp-matrix's LogPosDefDet() function
@ -825,13 +848,13 @@ template<class Real> static void UnitTestSherman() {
Matrix<Real> tt2(dimK, dimK); Matrix<Real> tt2(dimK, dimK);
tt2.AddMatMat(1.0, V, kTrans, tt1, kNoTrans, 0.0); tt2.AddMatMat(1.0, V, kTrans, tt1, kNoTrans, 0.0);
for (MatrixIndexT i = 0;i < dimK;i++) for (MatrixIndexT i = 0;i < dimK;i++)
for (MatrixIndexT j = 0;j < dimK;j++) for (MatrixIndexT j = 0;j < dimK;j++)
{ {
if (i == j) if (i == j)
I(i, j) = 1.0; I(i, j) = 1.0;
else else
I(i, j) = 0.0; I(i, j) = 0.0;
} }
tt2.AddMat(1.0, I); // I = identity tt2.AddMat(1.0, I); // I = identity
tt2.Invert(); tt2.Invert();
@ -975,25 +998,25 @@ template<class Real> static void UnitTestSvdNodestroy() {
/* /*
template<class Real> static void UnitTestSvdVariants() { // just make sure it doesn't crash if we call it but don't want left or right singular vectors. there are KALDI_ASSERTs inside the Svd. template<class Real> static void UnitTestSvdVariants() { // just make sure it doesn't crash if we call it but don't want left or right singular vectors. there are KALDI_ASSERTs inside the Svd.
#ifndef HAVE_ATLAS #ifndef HAVE_ATLAS
int Base = 10, Rand = 5, Iter = 25; int Base = 10, Rand = 5, Iter = 25;
for (int iter = 0;iter < Iter;iter++) { for (int iter = 0;iter < Iter;iter++) {
MatrixIndexT dimM = Base + rand() % Rand, dimN = Base + rand() % Rand; MatrixIndexT dimM = Base + rand() % Rand, dimN = Base + rand() % Rand;
// if (dimM<dimN) std::swap(dimM, dimN); // M>=N. // if (dimM<dimN) std::swap(dimM, dimN); // M>=N.
Matrix<Real> M(dimM, dimN); Matrix<Real> M(dimM, dimN);
Matrix<Real> U(dimM, dimM), Vt(dimN, dimN); Vector<Real> v(std::min(dimM, dimN)); Matrix<Real> U(dimM, dimM), Vt(dimN, dimN); Vector<Real> v(std::min(dimM, dimN));
Matrix<Real> Utmp(dimM, 1); Matrix<Real> Vttmp(1, dimN); Matrix<Real> Utmp(dimM, 1); Matrix<Real> Vttmp(1, dimN);
InitRand(&M); InitRand(&M);
M.Svd(v, U, Vttmp, "A", "N"); M.Svd(v, U, Vttmp, "A", "N");
M.Svd(v, Utmp, Vt, "N", "A"); M.Svd(v, Utmp, Vt, "N", "A");
Matrix<Real> U2(dimM, dimM), Vt2(dimN, dimN); Vector<Real> v2(std::min(dimM, dimN)); Matrix<Real> U2(dimM, dimM), Vt2(dimN, dimN); Vector<Real> v2(std::min(dimM, dimN));
M.Svd(v, U2, Vt2, "A", "A"); M.Svd(v, U2, Vt2, "A", "A");
AssertEqual(U, U2); AssertEqual(Vt, Vt2); AssertEqual(U, U2); AssertEqual(Vt, Vt2);
} }
#endif #endif
}*/ }*/
template<class Real> static void UnitTestSvdJustvec() { // Making sure gives same answer if we get just the vector, not the eigs. template<class Real> static void UnitTestSvdJustvec() { // Making sure gives same answer if we get just the vector, not the eigs.
int Base = 10, Rand = 5, Iter = 25; int Base = 10, Rand = 5, Iter = 25;
@ -1034,9 +1057,9 @@ template<class Real> static void UnitTestEig() {
for (int iter = 0;iter < 5;iter++) { for (int iter = 0;iter < 5;iter++) {
MatrixIndexT dimM = 1 + iter; MatrixIndexT dimM = 1 + iter;
/* if (iter < 10) /* if (iter < 10)
dimM = 1 + rand() % 6; dimM = 1 + rand() % 6;
else else
dimM = 5 + rand()%10; */ dimM = 5 + rand()%10; */
Matrix<Real> M(dimM, dimM); Matrix<Real> M(dimM, dimM);
InitRand(&M); InitRand(&M);
Matrix<Real> P(dimM, dimM); Matrix<Real> P(dimM, dimM);
@ -1392,21 +1415,21 @@ static void UnitTestTransposeScatter() {
for (MatrixIndexT i = 0;i < Ap.NumRows();i++) { for (MatrixIndexT i = 0;i < Ap.NumRows();i++) {
for (MatrixIndexT j = 0; j<=i; j++) { for (MatrixIndexT j = 0; j<=i; j++) {
Ap(i, j) = RandGauss(); Ap(i, j) = RandGauss();
} }
} }
for (MatrixIndexT i = 0;i < M.NumRows();i++) { for (MatrixIndexT i = 0;i < M.NumRows();i++) {
for (MatrixIndexT j = 0; j < M.NumCols(); j++) { for (MatrixIndexT j = 0; j < M.NumCols(); j++) {
M(i, j) = RandGauss(); M(i, j) = RandGauss();
} }
} }
/* /*
std::stringstream ss("1 2 3"); std::stringstream ss("1 2 3");
ss >> Ap; ss >> Ap;
ss.clear(); ss.clear();
ss.str("5 6 7 8 9 10"); ss.str("5 6 7 8 9 10");
ss >> M; ss >> M;
*/ */
Af.CopyFromSp(Ap); Af.CopyFromSp(Ap);
A_MT.AddMatMat(1.0, Af, kNoTrans, M, kTrans, 0.0); A_MT.AddMatMat(1.0, Af, kNoTrans, M, kTrans, 0.0);
@ -1414,9 +1437,9 @@ static void UnitTestTransposeScatter() {
Op.AddMat2Sp(1.0, M, kNoTrans, Ap, 0.0); Op.AddMat2Sp(1.0, M, kNoTrans, Ap, 0.0);
// std::cout << "A" << '\n' << Af << '\n'; // std::cout << "A" << '\n' << Af << '\n';
// std::cout << "M" << '\n' << M << '\n'; // std::cout << "M" << '\n' << M << '\n';
// std::cout << "Op" << '\n' << Op << '\n'; // std::cout << "Op" << '\n' << Op << '\n';
for (MatrixIndexT i = 0; i < dimO; i++) { for (MatrixIndexT i = 0; i < dimO; i++) {
for (MatrixIndexT j = 0; j<=i; j++) { for (MatrixIndexT j = 0; j<=i; j++) {
@ -1429,8 +1452,8 @@ static void UnitTestTransposeScatter() {
Af.AddMatMat(1.0, M, kTrans, A_MT, kNoTrans, 1.0); Af.AddMatMat(1.0, M, kTrans, A_MT, kNoTrans, 1.0);
Ap.AddMat2Sp(1.0, M, kTrans, Op, 1.0); Ap.AddMat2Sp(1.0, M, kTrans, Op, 1.0);
// std::cout << "Ap" << '\n' << Ap << '\n'; // std::cout << "Ap" << '\n' << Ap << '\n';
// std::cout << "Af" << '\n' << Af << '\n'; // std::cout << "Af" << '\n' << Af << '\n';
for (MatrixIndexT i = 0; i < dimA; i++) { for (MatrixIndexT i = 0; i < dimA; i++) {
for (MatrixIndexT j = 0; j<=i; j++) { for (MatrixIndexT j = 0; j<=i; j++) {
@ -1671,7 +1694,7 @@ template<class Real> static void UnitTestSimple() {
S.CopyFromMat(N); S.CopyFromMat(N);
KALDI_ASSERT(S.IsZero()); KALDI_ASSERT(S.IsZero());
KALDI_ASSERT(S.IsDiagonal()); KALDI_ASSERT(S.IsDiagonal());
} }
} }
@ -1885,14 +1908,14 @@ template<class Real> static void UnitTestRange() { // Testing SubMatrix class.
KALDI_ASSERT(sub.Sum() == V.Range(lenStart, lenEnd-lenStart).Sum()); KALDI_ASSERT(sub.Sum() == V.Range(lenStart, lenEnd-lenStart).Sum());
for (MatrixIndexT i = lenStart;i < lenEnd;i++) for (MatrixIndexT i = lenStart;i < lenEnd;i++)
KALDI_ASSERT(V(i) == sub(i-lenStart)); KALDI_ASSERT(V(i) == sub(i-lenStart));
InitRand(&sub); InitRand(&sub);
KALDI_ASSERT(sub.Sum() == V.Range(lenStart, lenEnd-lenStart).Sum()); KALDI_ASSERT(sub.Sum() == V.Range(lenStart, lenEnd-lenStart).Sum());
for (MatrixIndexT i = lenStart;i < lenEnd;i++) for (MatrixIndexT i = lenStart;i < lenEnd;i++)
KALDI_ASSERT(V(i) == sub(i-lenStart)); KALDI_ASSERT(V(i) == sub(i-lenStart));
} }
} }
@ -2047,7 +2070,7 @@ template<class Real> static void UnitTestSolve() {
SolveQuadraticProblem(H, g, &x2); SolveQuadraticProblem(H, g, &x2);
#endif #endif
KALDI_ASSERT(VecVec(x2, g) -0.5* VecSpVec(x2, H, x2) >= KALDI_ASSERT(VecVec(x2, g) -0.5* VecSpVec(x2, H, x2) >=
VecVec(x, g) -0.5* VecSpVec(x, H, x)); VecVec(x, g) -0.5* VecSpVec(x, H, x));
// Check objf not decreased. // Check objf not decreased.
} }
@ -2086,16 +2109,16 @@ template<class Real> static void UnitTestSolve() {
M3.AddMatSp(1.0, Y, kNoTrans, Qinv, 0.0); M3.AddMatSp(1.0, Y, kNoTrans, Qinv, 0.0);
if (Q.Cond() < 1000.0) { if (Q.Cond() < 1000.0) {
AssertEqual(M2, M3); // This equality only holds if SigmaInv full-rank, AssertEqual(M2, M3); // This equality only holds if SigmaInv full-rank,
// which is overwhelmingly likely if dimO > dimM // which is overwhelmingly likely if dimO > dimM
} }
{ {
Real a1 = TraceMatSpMat(M2, kTrans, SigmaInv, Y, kNoTrans), Real a1 = TraceMatSpMat(M2, kTrans, SigmaInv, Y, kNoTrans),
a2 = TraceMatSpMatSp(M2, kNoTrans, Q, M2, kTrans, SigmaInv), a2 = TraceMatSpMatSp(M2, kNoTrans, Q, M2, kTrans, SigmaInv),
b1 = TraceMatSpMat(M, kTrans, SigmaInv, Y, kNoTrans), b1 = TraceMatSpMat(M, kTrans, SigmaInv, Y, kNoTrans),
b2 = TraceMatSpMatSp(M, kNoTrans, Q, M, kTrans, SigmaInv), b2 = TraceMatSpMatSp(M, kNoTrans, Q, M, kTrans, SigmaInv),
a3 = a1 - 0.5 * a2, a3 = a1 - 0.5 * a2,
b3 = b1 - 0.5 * b2; b3 = b1 - 0.5 * b2;
KALDI_ASSERT(a3 >= b3); KALDI_ASSERT(a3 >= b3);
// KALDI_LOG << "a3 = " << a3 << ", b3 = " << b3 << ", c3 = " << c3; // KALDI_LOG << "a3 = " << a3 << ", b3 = " << b3 << ", c3 = " << c3;
} // Check objf not decreased. } // Check objf not decreased.
@ -2114,8 +2137,8 @@ template<class Real> static void UnitTestSolve() {
Matrix<Real> M(dimO, dimM), G(dimO, dimM); Matrix<Real> M(dimO, dimM), G(dimO, dimM);
M.SetRandn(); M.SetRandn();
G.SetRandn(); G.SetRandn();
// InitRand(&M); // InitRand(&M);
// InitRand(&G); // InitRand(&G);
Matrix<Real> M2(M); Matrix<Real> M2(M);
@ -2128,13 +2151,13 @@ template<class Real> static void UnitTestSolve() {
{ {
Real a1 = TraceMatMat(M2, G, kTrans), Real a1 = TraceMatMat(M2, G, kTrans),
a2 = TraceMatSpMatSp(M2, kNoTrans, Q1, M2, kTrans, P1), a2 = TraceMatSpMatSp(M2, kNoTrans, Q1, M2, kTrans, P1),
a3 = TraceMatSpMatSp(M2, kNoTrans, Q2, M2, kTrans, P2), a3 = TraceMatSpMatSp(M2, kNoTrans, Q2, M2, kTrans, P2),
b1 = TraceMatMat(M, G, kTrans), b1 = TraceMatMat(M, G, kTrans),
b2 = TraceMatSpMatSp(M, kNoTrans, Q1, M, kTrans, P1), b2 = TraceMatSpMatSp(M, kNoTrans, Q1, M, kTrans, P1),
b3 = TraceMatSpMatSp(M, kNoTrans, Q2, M, kTrans, P2), b3 = TraceMatSpMatSp(M, kNoTrans, Q2, M, kTrans, P2),
a4 = a1 - 0.5 * a2 - 0.5 * a3, a4 = a1 - 0.5 * a2 - 0.5 * a3,
b4 = b1 - 0.5 * b2 - 0.5 * b3; b4 = b1 - 0.5 * b2 - 0.5 * b3;
KALDI_LOG << "a4 = " << a4 << ", b4 = " << b4; KALDI_LOG << "a4 = " << a4 << ", b4 = " << b4;
KALDI_ASSERT(a4 >= b4); KALDI_ASSERT(a4 >= b4);
} // Check objf not decreased. } // Check objf not decreased.
@ -2205,15 +2228,15 @@ template<class Real> static void UnitTestTrace() {
ABC.AddMatMat(1.0, A, kNoTrans, BC, kNoTrans, 0.0); ABC.AddMatMat(1.0, A, kNoTrans, BC, kNoTrans, 0.0);
Real Real
t1 = TraceMat(ABC), t1 = TraceMat(ABC),
t2 = ABC.Trace(), t2 = ABC.Trace(),
t3 = TraceMatMat(A, BC), t3 = TraceMatMat(A, BC),
t4 = TraceMatMat(AT, BC, kTrans), t4 = TraceMatMat(AT, BC, kTrans),
t5 = TraceMatMat(BC, AT, kTrans), t5 = TraceMatMat(BC, AT, kTrans),
t6 = TraceMatMatMat(A, kNoTrans, B, kNoTrans, C, kNoTrans), t6 = TraceMatMatMat(A, kNoTrans, B, kNoTrans, C, kNoTrans),
t7 = TraceMatMatMat(AT, kTrans, B, kNoTrans, C, kNoTrans), t7 = TraceMatMatMat(AT, kTrans, B, kNoTrans, C, kNoTrans),
t8 = TraceMatMatMat(AT, kTrans, BT, kTrans, C, kNoTrans), t8 = TraceMatMatMat(AT, kTrans, BT, kTrans, C, kNoTrans),
t9 = TraceMatMatMat(AT, kTrans, BT, kTrans, CT, kTrans); t9 = TraceMatMatMat(AT, kTrans, BT, kTrans, CT, kTrans);
Matrix<Real> ABC1(dimM, dimP); // tests AddMatMatMat. Matrix<Real> ABC1(dimM, dimP); // tests AddMatMatMat.
ABC1.AddMatMatMat(1.0, A, kNoTrans, B, kNoTrans, C, kNoTrans, 0.0); ABC1.AddMatMatMat(1.0, A, kNoTrans, B, kNoTrans, C, kNoTrans, 0.0);
@ -2228,8 +2251,8 @@ template<class Real> static void UnitTestTrace() {
Real tol = 0.001; Real tol = 0.001;
KALDI_ASSERT((std::abs(t1-t2) < tol) && (std::abs(t2-t3) < tol) && (std::abs(t3-t4) < tol) KALDI_ASSERT((std::abs(t1-t2) < tol) && (std::abs(t2-t3) < tol) && (std::abs(t3-t4) < tol)
&& (std::abs(t4-t5) < tol) && (std::abs(t5-t6) < tol) && (std::abs(t6-t7) < tol) && (std::abs(t4-t5) < tol) && (std::abs(t5-t6) < tol) && (std::abs(t6-t7) < tol)
&& (std::abs(t7-t8) < tol) && (std::abs(t8-t9) < tol)); && (std::abs(t7-t8) < tol) && (std::abs(t8-t9) < tol));
} }
for (MatrixIndexT i = 0;i < 5;i++) { for (MatrixIndexT i = 0;i < 5;i++) {
@ -2833,7 +2856,7 @@ template<class Real> static void UnitTestCompressedMatrix() {
for (int i = 0; i < num_rows; i++) { for (int i = 0; i < num_rows; i++) {
Vector<Real> V(num_cols); Vector<Real> V(num_cols);
cmat.CopyRowToVec(i, &V); // get row. cmat.CopyRowToVec(i, &V); // get row.
for (MatrixIndexT k = 0;k < num_cols;k++) { for (MatrixIndexT k = 0; k < num_cols; k++) {
AssertEqual(M2(i, k), V(k)); AssertEqual(M2(i, k), V(k));
} }
} }
@ -2850,10 +2873,10 @@ template<class Real> static void UnitTestCompressedMatrix() {
//test of getting a submatrix //test of getting a submatrix
if(num_rows != 0 && num_cols != 0){ if(num_rows != 0 && num_cols != 0){
int32 sub_row_offset = (num_rows == 1 ? 0 : rand() % (num_rows-1)), int32 sub_row_offset = (num_rows == 1 ? 0 : rand() % (num_rows-1)),
sub_col_offset = (num_cols == 1 ? 0 : rand() % (num_cols-1)); sub_col_offset = (num_cols == 1 ? 0 : rand() % (num_cols-1));
// to make sure we don't mod by zero // to make sure we don't mod by zero
int32 num_subrows = rand() % (num_rows-sub_row_offset), int32 num_subrows = rand() % (num_rows-sub_row_offset),
num_subcols = rand() % (num_cols-sub_col_offset); num_subcols = rand() % (num_cols-sub_col_offset);
if(num_subrows == 0 || num_subcols == 0){ // in case we randomized to if(num_subrows == 0 || num_subcols == 0){ // in case we randomized to
// empty matrix, at least make it correct // empty matrix, at least make it correct
num_subrows = 0; num_subrows = 0;
@ -2926,7 +2949,7 @@ template<class Real> static void MatrixUnitTest() {
UnitTestSplitRadixComplexFft2<Real>(); UnitTestSplitRadixComplexFft2<Real>();
UnitTestDct<Real>(); UnitTestDct<Real>();
UnitTestRealFft<Real>(); UnitTestRealFft<Real>();
KALDI_LOG << " Point C"; KALDI_LOG << " Point C";
UnitTestSplitRadixRealFft<Real>(); UnitTestSplitRadixRealFft<Real>();
UnitTestSvd<Real>(); UnitTestSvd<Real>();
UnitTestSvdNodestroy<Real>(); UnitTestSvdNodestroy<Real>();
@ -2934,14 +2957,14 @@ template<class Real> static void MatrixUnitTest() {
UnitTestSpAddVec<Real, float>(); UnitTestSpAddVec<Real, float>();
UnitTestSpAddVec<Real, double>(); UnitTestSpAddVec<Real, double>();
UnitTestSpInvert<Real>(); UnitTestSpInvert<Real>();
KALDI_LOG << " Point D"; KALDI_LOG << " Point D";
UnitTestTpInvert<Real>(); UnitTestTpInvert<Real>();
UnitTestIo<Real>(); UnitTestIo<Real>();
UnitTestIoCross<Real>(); UnitTestIoCross<Real>();
UnitTestHtkIo<Real>(); UnitTestHtkIo<Real>();
UnitTestScale<Real>(); UnitTestScale<Real>();
UnitTestTrace<Real>(); UnitTestTrace<Real>();
KALDI_LOG << " Point E"; KALDI_LOG << " Point E";
CholeskyUnitTestTr<Real>(); CholeskyUnitTestTr<Real>();
UnitTestAxpy<Real>(); UnitTestAxpy<Real>();
UnitTestSimple<Real>(); UnitTestSimple<Real>();
@ -2954,7 +2977,7 @@ template<class Real> static void MatrixUnitTest() {
// UnitTestSvdVariants<Real>(); // UnitTestSvdVariants<Real>();
UnitTestPower<Real>(); UnitTestPower<Real>();
UnitTestDeterminant<Real>(); UnitTestDeterminant<Real>();
KALDI_LOG << " Point F"; KALDI_LOG << " Point F";
UnitTestDeterminantSign<Real>(); UnitTestDeterminantSign<Real>();
UnitTestSger<Real>(); UnitTestSger<Real>();
UnitTestAddOuterProductPlusMinus<Real>(); UnitTestAddOuterProductPlusMinus<Real>();
@ -2965,13 +2988,14 @@ template<class Real> static void MatrixUnitTest() {
UnitTestSherman<Real>(); UnitTestSherman<Real>();
UnitTestSpVec<Real>(); UnitTestSpVec<Real>();
UnitTestLimitCondInvert<Real>(); UnitTestLimitCondInvert<Real>();
KALDI_LOG << " Point G"; KALDI_LOG << " Point G";
UnitTestFloorChol<Real>(); UnitTestFloorChol<Real>();
UnitTestFloorUnit<Real>(); UnitTestFloorUnit<Real>();
UnitTestLimitCond<Real>(); UnitTestLimitCond<Real>();
UnitTestMat2Vec<Real>(); UnitTestMat2Vec<Real>();
UnitTestSpLogExp<Real>(); UnitTestSpLogExp<Real>();
KALDI_LOG << " Point H"; KALDI_LOG << " Point H";
UnitTestCopyRowsAndCols<Real>();
UnitTestSpliceRows<Real>(); UnitTestSpliceRows<Real>();
UnitTestAddSp<Real>(); UnitTestAddSp<Real>();
UnitTestRemoveRow<Real>(); UnitTestRemoveRow<Real>();
@ -2982,12 +3006,12 @@ template<class Real> static void MatrixUnitTest() {
UnitTestSimpleForMat<Real>(); UnitTestSimpleForMat<Real>();
UnitTestNorm<Real>(); UnitTestNorm<Real>();
UnitTestMul<Real>(); UnitTestMul<Real>();
KALDI_LOG << " Point I"; KALDI_LOG << " Point I";
UnitTestSolve<Real>(); UnitTestSolve<Real>();
UnitTestMaxMin<Real>(); UnitTestMaxMin<Real>();
UnitTestInnerProd<Real>(); UnitTestInnerProd<Real>();
UnitTestScaleDiag<Real>(); UnitTestScaleDiag<Real>();
KALDI_LOG << " Point J"; KALDI_LOG << " Point J";
UnitTestTraceSpSpLower<Real>(); UnitTestTraceSpSpLower<Real>();
UnitTestTranspose<Real>(); UnitTestTranspose<Real>();
UnitTestAddVecToRows<Real>(); UnitTestAddVecToRows<Real>();

Просмотреть файл

@ -72,7 +72,7 @@ class BiasedLinearity : public UpdatableComponent {
// compute gradient // compute gradient
linearity_corr_.AddMatMat(1.0, err, kTrans, input, kNoTrans, momentum_); linearity_corr_.AddMatMat(1.0, err, kTrans, input, kNoTrans, momentum_);
bias_corr_.Scale(momentum_); bias_corr_.Scale(momentum_);
bias_corr_.AddRowSumMat(err); bias_corr_.AddRowSumMat(1.0, err);
// l2 regularization // l2 regularization
if (l2_penalty_ != 0.0) { if (l2_penalty_ != 0.0) {
linearity_.AddMat(-learn_rate_*l2_penalty_*input.NumRows(), linearity_); linearity_.AddMat(-learn_rate_*l2_penalty_*input.NumRows(), linearity_);

Просмотреть файл

@ -128,7 +128,7 @@ void Rnnlm::Backpropagate(const MatrixBase<BaseFloat> &in_err) {
// update layer2 // update layer2
W2_.AddMatMat(-learn_rate_, h2_, kTrans, in_err, kNoTrans, 1.0); W2_.AddMatMat(-learn_rate_, h2_, kTrans, in_err, kNoTrans, 1.0);
b2_corr_.SetZero(); b2_corr_.SetZero();
b2_corr_.AddRowSumMat(in_err); b2_corr_.AddRowSumMat(1.0, in_err);
b2_.AddVec(-learn_rate_, b2_corr_); b2_.AddVec(-learn_rate_, b2_corr_);
// LAYER1 // LAYER1
@ -136,7 +136,7 @@ void Rnnlm::Backpropagate(const MatrixBase<BaseFloat> &in_err) {
U1_corr_.SetZero(); U1_corr_.SetZero();
b1_corr_.SetZero(); b1_corr_.SetZero();
// accumulate gradient for layer1 // accumulate gradient for layer1
b1_corr_.AddRowSumMat(e2_); b1_corr_.AddRowSumMat(1.0, e2_);
for(int32 r=0; r<e2_.NumRows();r++) { for(int32 r=0; r<e2_.NumRows();r++) {
V1_corr_.Row(in_seq_[r]-1).AddVec(1.0, e2_.Row(r)); V1_corr_.Row(in_seq_[r]-1).AddVec(1.0, e2_.Row(r));
} }
@ -174,7 +174,7 @@ void Rnnlm::Backpropagate(const MatrixBase<BaseFloat> &in_err) {
} }
// accumulate graidient // accumulate graidient
b1_corr_.AddRowSumMat(E); b1_corr_.AddRowSumMat(1.0, E);
for(int32 r=0; r<E.NumRows();r++) { for(int32 r=0; r<E.NumRows();r++) {
// :TODO: IS IT CORRECT? // :TODO: IS IT CORRECT?
// V1_corr_.Row(in_seq_[r+step]).AddVec(1.0,E.Row(r)); // V1_corr_.Row(in_seq_[r+step]).AddVec(1.0,E.Row(r));

Просмотреть файл

@ -273,8 +273,8 @@ double EbwAmSgmmUpdater::UpdateM(const MleAmSgmmAccs &num_accs,
Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I); Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I);
for (int32 j = 0; j < num_accs.num_states_; j++) { for (int32 j = 0; j < num_accs.num_states_; j++) {
num_count_vec.AddRowSumMat(num_accs.gamma_[j]); num_count_vec.AddRowSumMat(1.0, num_accs.gamma_[j]);
den_count_vec.AddRowSumMat(den_accs.gamma_[j]); den_count_vec.AddRowSumMat(1.0, den_accs.gamma_[j]);
} }
for (int32 i = 0; i < I; i++) { for (int32 i = 0; i < I; i++) {
@ -363,8 +363,8 @@ double EbwAmSgmmUpdater::UpdateWParallel(const MleAmSgmmAccs &num_accs,
Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I); Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I);
for (int32 j = 0; j < num_accs.num_states_; j++) { for (int32 j = 0; j < num_accs.num_states_; j++) {
num_count_vec.AddRowSumMat(num_accs.gamma_[j]); num_count_vec.AddRowSumMat(1.0, num_accs.gamma_[j]);
den_count_vec.AddRowSumMat(den_accs.gamma_[j]); den_count_vec.AddRowSumMat(1.0, den_accs.gamma_[j]);
} }
// Get the F_i and g_i quantities-- this is done in parallel (multi-core), // Get the F_i and g_i quantities-- this is done in parallel (multi-core),
@ -448,8 +448,8 @@ double EbwAmSgmmUpdater::UpdateN(const MleAmSgmmAccs &num_accs,
Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I); Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I);
for (int32 j = 0; j < num_accs.num_states_; j++) { for (int32 j = 0; j < num_accs.num_states_; j++) {
num_count_vec.AddRowSumMat(num_accs.gamma_[j]); num_count_vec.AddRowSumMat(1.0, num_accs.gamma_[j]);
den_count_vec.AddRowSumMat(den_accs.gamma_[j]); den_count_vec.AddRowSumMat(1.0, den_accs.gamma_[j]);
} }
for (int32 i = 0; i < I; i++) { for (int32 i = 0; i < I; i++) {
@ -522,8 +522,8 @@ double EbwAmSgmmUpdater::UpdateVars(const MleAmSgmmAccs &num_accs,
Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I); Vector<double> num_count_vec(I), den_count_vec(I), impr_vec(I);
for (int32 j = 0; j < num_accs.num_states_; j++) { for (int32 j = 0; j < num_accs.num_states_; j++) {
num_count_vec.AddRowSumMat(num_accs.gamma_[j]); num_count_vec.AddRowSumMat(1.0, num_accs.gamma_[j]);
den_count_vec.AddRowSumMat(den_accs.gamma_[j]); den_count_vec.AddRowSumMat(1.0, den_accs.gamma_[j]);
} }
for (int32 i = 0; i < I; i++) { for (int32 i = 0; i < I; i++) {

Просмотреть файл

@ -538,7 +538,7 @@ BaseFloat AmSgmm2::LogLikelihood(const Sgmm2PerFrameDerivedVars &per_frame_vars,
substate_cache.remaining_log_like = max; substate_cache.remaining_log_like = max;
int32 num_substates = loglikes.NumCols(); int32 num_substates = loglikes.NumCols();
substate_cache.likes.Resize(num_substates); // zeroes it. substate_cache.likes.Resize(num_substates); // zeroes it.
substate_cache.likes.AddRowSumMat(loglikes); // add likelihoods [not in log!] for substate_cache.likes.AddRowSumMat(1.0, loglikes); // add likelihoods [not in log!] for
// each column [i.e. summing over the rows], so we get the sum for // each column [i.e. summing over the rows], so we get the sum for
// each substate index. You have to multiply by exp(remaining_log_like) // each substate index. You have to multiply by exp(remaining_log_like)
// to get a real likelihood. // to get a real likelihood.
@ -613,7 +613,7 @@ void AmSgmm2::SplitSubstatesInGroup(const Vector<BaseFloat> &pdf_occupancies,
int32 split_m; // substate to split. int32 split_m; // substate to split.
{ {
Vector<BaseFloat> substate_count(tgt_M); Vector<BaseFloat> substate_count(tgt_M);
substate_count.AddRowSumMat(c_j); substate_count.AddRowSumMat(1.0, c_j);
BaseFloat *data = substate_count.Data(); BaseFloat *data = substate_count.Data();
split_m = std::max_element(data, data+cur_M) - data; split_m = std::max_element(data, data+cur_M) - data;
} }

Просмотреть файл

@ -50,10 +50,10 @@ void EbwAmSgmm2Updater::Update(const MleAmSgmm2Accs &num_accs,
Vector<double> gamma_num(num_accs.num_gaussians_); Vector<double> gamma_num(num_accs.num_gaussians_);
for (int32 j1 = 0; j1 < num_accs.num_groups_; j1++) for (int32 j1 = 0; j1 < num_accs.num_groups_; j1++)
gamma_num.AddRowSumMat(num_accs.gamma_[j1]); gamma_num.AddRowSumMat(1.0, num_accs.gamma_[j1]);
Vector<double> gamma_den(den_accs.num_gaussians_); Vector<double> gamma_den(den_accs.num_gaussians_);
for (int32 j1 = 0; j1 < den_accs.num_groups_; j1++) for (int32 j1 = 0; j1 < den_accs.num_groups_; j1++)
gamma_den.AddRowSumMat(den_accs.gamma_[j1]); gamma_den.AddRowSumMat(1.0, den_accs.gamma_[j1]);
BaseFloat tot_impr = 0.0; BaseFloat tot_impr = 0.0;

Просмотреть файл

@ -617,7 +617,7 @@ void MleAmSgmm2Updater::Update(const MleAmSgmm2Accs &accs,
Vector<double> gamma_i(accs.num_gaussians_); Vector<double> gamma_i(accs.num_gaussians_);
for (int32 j1 = 0; j1 < accs.num_groups_; j1++) for (int32 j1 = 0; j1 < accs.num_groups_; j1++)
gamma_i.AddRowSumMat(accs.gamma_[j1]); // add sum of rows of gamma_i.AddRowSumMat(1.0, accs.gamma_[j1]); // add sum of rows of
// accs.gamma_[j1], to gamma_i. // accs.gamma_[j1], to gamma_i.
if (flags & kSgmmPhoneProjections) if (flags & kSgmmPhoneProjections)

Просмотреть файл

@ -63,7 +63,7 @@ void LdaEstimate::Estimate(int32 target_dim,
// total covariance // total covariance
double sum = zero_acc_.Sum(); double sum = zero_acc_.Sum();
Vector<double> total_mean(dim); Vector<double> total_mean(dim);
total_mean.AddRowSumMat(first_acc_); total_mean.AddRowSumMat(1.0, first_acc_);
total_mean.Scale(1/sum); total_mean.Scale(1/sum);
SpMatrix<double> total_covar(total_second_acc_); SpMatrix<double> total_covar(total_second_acc_);
total_covar.Scale(1/sum); total_covar.Scale(1/sum);