add the expand copy nnet trasforms, option randomize to nnet trainer

git-svn-id: https://svn.code.sf.net/p/kaldi/code/trunk@1197 5e6a8d80-dfce-4ca6-a32a-6e07a63d50c8

src/nnet/nnet-component.cc

@@ -21,6 +21,7 @@
 #include "nnet/nnet-activation.h"
 #include "nnet/nnet-biasedlinearity.h"
 #include "nnet/nnet-rbm.h"
+#include "nnet/nnet-various.h"
 
 namespace kaldi {
 
@@ -83,6 +84,12 @@ Component* Component::Read(std::istream &is, bool binary, Nnet *nnet) {
     case Component::kRbm :
       p_comp = new Rbm(dim_in, dim_out, nnet);
       break;
+    case Component::kExpand :
+      p_comp = new Expand(dim_in, dim_out, nnet);
+      break;
+    case Component::kCopy :
+      p_comp = new Copy(dim_in, dim_out, nnet);
+      break;
     case Component::kUnknown :
     default :
       KALDI_ERR << "Missing type: " << token;

src/nnet/nnet-various.h

@@ -57,7 +57,7 @@ class Expand : public Component {
     frame_offsets_.CopyFromVec(vec_i); 
   }
 
-  void WriteData(std::ostream &os, bool binary) {
+  void WriteData(std::ostream &os, bool binary) const {
     std::vector<int32> vec_i;
     frame_offsets_.CopyToVec(&vec_i);
     Vector<double> vec_d(vec_i.size());
@@ -111,7 +111,7 @@ class Copy : public Component {
     copy_from_indices_.CopyFromVec(vec_i); 
   }
 
-  void WriteData(std::ostream &os, bool binary) { 
+  void WriteData(std::ostream &os, bool binary) const { 
     std::vector<int32> vec_i;
     copy_from_indices_.CopyToVec(&vec_i);
     Vector<double> vec_d(vec_i.size());

src/nnetbin/nnet-train-xent-hardlab-frmshuff.cc

@@ -35,9 +35,11 @@ int main(int argc, char *argv[]) {
 
     ParseOptions po(usage);
     bool binary = false, 
-         crossvalidate = false;
+         crossvalidate = false,
+         randomize = true;
     po.Register("binary", &binary, "Write output in binary mode");
     po.Register("cross-validate", &crossvalidate, "Perform cross-validation (don't backpropagate)");
+    po.Register("randomize", &randomize, "Perform the frame-level shuffling within the Cache::");
 
     BaseFloat learn_rate = 0.008,
         momentum = 0.0,
@@ -139,7 +141,7 @@ int main(int argc, char *argv[]) {
         time_next += t_features.Elapsed();
       }
       // randomize
-      if (!crossvalidate) {
+      if (!crossvalidate && randomize) {
         cache.Randomize();
       }
       // report