example_set.cpp

支持向量机(SVM)的VC源代码

void example_set_c::output_ys(ostream& data_stream) const{
  data_stream<<"# examples ys"<<endl;
  for(SVMINT i=0;i<examples_total;i++){
    data_stream<<(the_set[i].y)<<endl;
  };  
};


void readnext(istream& i, char* s, const char delimiter){
  SVMINT pos=0;
  char next = i.peek();
  // skip whitespace
  while((! i.eof()) &&
	(('\n' == next) ||
	 (' ' == next)  ||
	 ('\t' == next) ||
	 ('\r' == next) ||
	 ('\f' == next))){
    i.get();
    next = i.peek();
  };
  // read next token
  if(delimiter == next){
    s[pos] = '0';
    pos++;
    next = i.peek();
  }
  else{
    while((! i.eof()) &&
	  ('\n' != next) &&
	  (' ' != next) &&
	  ('\t' != next) &&
	  ('\r' != next) &&
	  ('\f' != next) &&
	  (delimiter != next) &&
	  (pos < MAXCHAR-1)){
      s[pos] = i.get();
      pos++;
      next = i.peek();
    };
  };
  s[pos] = '\0';
  if(! (i.eof() || ('\n' == next))){
    // remove delimiter
    i.get();
  };
};


istream& operator>> (istream& data_stream, example_set_c& examples){
  // lower case, scale (y/n)
  char* s = new char[MAXCHAR]; // next item in the stream
  char* s2 = new char[MAXCHAR];
  long count=0; // number of examples read (does not necessarily equal examples_total
  char next=0; // first character in the stream
  char delimiter = examples.my_format.delimiter; // By which character are the numbers separated?
  int sparse = examples.my_format.sparse;
  int where_x = examples.my_format.where_x;  // format of the file
  int where_y = examples.my_format.where_y;
  int where_alpha = examples.my_format.where_alpha;
  SVMINT i;
  SVMINT given_total = 0; // what does the user say is the total of examples?
  SVMINT pos; // dummy for pos of attribute in example
  SVMINT dim = examples.get_dim();
  SVMFLOAT* new_example = new SVMFLOAT[dim+2]; // examples to be inserted

  while(('@' != next) && (! data_stream.eof())){
    try{
      next = data_stream.peek();
      if(('@' == next) || (data_stream.eof())){
	// end of this section
      }
      else if(('\n' == next) ||
	      (' ' == next) ||
	      ('\r' == next) ||
	      ('\f' == next) ||
	      ('\t' == next)){
	// ignore
	next = data_stream.get();
      }
      else if('#' == next){
	// line contains commentary
	data_stream.getline(s,MAXCHAR);
      }
      else if(('+' == next) || ('-' == next) || 
	      ('y' == next) || ('a' == next) ||
	      ((next >= '0') && (next <= '9'))){
	// read an example
	pos = 0;
	new_example[dim] = 0;
	new_example[dim+1] = 0;
	if(sparse){
	  for(pos=0;pos<dim;pos++){
	    new_example[pos] = 0;
	  };
	  while((! data_stream.eof()) && ('\n' != data_stream.peek())){
	    readnext(data_stream,s,delimiter);
	    SVMINT spos = 0;
	    while((s[spos] != '\0') && (s[spos] != ':')){
	      spos++;
	    };
	    if(s[spos] == '\0'){
	      // read y
	      try{
		new_example[dim] = string2svmfloat(s);
	      }
	      catch(...){
		throw read_exception("Class is no number - could not read example");
	      };
	      examples.set_initialised_y();
	    }
	    else{
	      if(s[spos-1] == 'a'){
		// read alpha
		strncpy(s2,s+spos+1,MAXCHAR-spos);
		try{
		  new_example[dim+1] = string2svmfloat(s2);
		}
		catch(...){
		  throw read_exception("Alpha is no number - could not read example");
		};
		examples.set_initialised_alpha();
	      }
	      else if(s[spos-1] == 'y'){
		// read y
		strncpy(s2,s+spos+1,MAXCHAR-spos);
		try{
		  new_example[dim] = string2svmfloat(s2);
		}
		catch(...){
		  throw read_exception("Class is no number - could not read example");
		};
		examples.set_initialised_y();
	      }
	      else{
		// input index runs from 1 to dim (svmlight-compatibility):
		pos = atoi(s); 
		if(pos <= 0){
		  throw read_exception("Index number not positive.");
		};
		if(pos>dim){
		  // raise dimension
		  examples.set_dim(pos);  
		  SVMFLOAT* example_dummy = new SVMFLOAT[pos+2];
		  example_dummy[pos] = new_example[dim];
		  example_dummy[pos+1] = new_example[dim+1];
		  for(i=0;i<dim;i++){
		    example_dummy[i] = new_example[i];
		  };
		  for(i=dim;i<pos;i++){
		    example_dummy[i] = 0;
		  };
		  dim = pos;
		  delete []new_example;
		  new_example = example_dummy;
		};
		try{
		  new_example[pos-1] = string2svmfloat(s+spos+1);
		}
		catch(...){
		  throw read_exception("Attribute is no number - could not read example");
		}; 
	      };
	    };
	    while((! data_stream.eof()) && 
		  ((' ' == data_stream.peek()) || 
		   ('\t' == data_stream.peek()))){
	      data_stream.get();
	    };
	  };
	  pos = dim; // mark as ok
	}
	else{
	  // not sparse
	  for(int i=1;i<=3;i++) {
	    if(i ==  where_x){
	      // read attributes
	      if(dim <= 0) {
		// read & get dim
		char next_ws = data_stream.peek();
		dim=0;
		pos = 0;
		while(!(data_stream.eof() || ('\n' == next_ws))){
		  // try to read another attribute
		  while((! data_stream.eof()) &&
			((' ' == next_ws) ||
			 ('\t' == next_ws))){
		    data_stream.get();
		    next_ws = data_stream.peek();
		  };
		  if(!(data_stream.eof() || ('\n' == next_ws))){
		    // attribute is there, read it
		    if(pos == dim){
		      // double dim
		      dim = 2*dim+1;
		      SVMFLOAT* dummy = new_example;
		      new_example = new SVMFLOAT[dim+2];
		      new_example[dim] = dummy[pos];
		      new_example[dim+1] = dummy[pos+1];
		      for(SVMINT j=0;j<pos;j++){
			new_example[j] = dummy[j];
		      };
		      delete []dummy;
		    };
		    // read example into act_pos
		    readnext(data_stream,s,delimiter);
		    try{
		      new_example[pos]= string2svmfloat(s);
		    }
		    catch(...){
		      throw read_exception("Attribute is no number - could not read example");
		    };
		    pos++;
		    next_ws = data_stream.peek();
		  };
		};
		// line finished, set dim and exit
		if(where_y > where_x){
		  pos--;
		  // y at pos or pos+1 (one of xya xay xy)
		  if(where_y < where_alpha){
		    // xya
		    pos--;
		    new_example[dim] = new_example[pos];
		    new_example[dim+1] = new_example[pos+1];
		  }
		  else if(where_alpha < where_x){
		    // xy
		    new_example[dim] = new_example[pos];
		  }
		  else{
		    // xay
		    pos--;
		    SVMFLOAT dummy = new_example[pos]; // if pos==dim
		    new_example[dim] = new_example[pos+1];
		    new_example[dim+1] = dummy;
		  };
		}
		else if(where_alpha > where_x){
		  // xa
		  pos--;
		  new_example[dim+1] = new_example[pos];
		};
		SVMFLOAT* dummy = new_example;
		new_example = new SVMFLOAT[pos+2];
		for(SVMINT j=0;j<pos;j++){
		  new_example[j] = dummy[j];
		};
		new_example[pos] = dummy[dim];
		new_example[pos+1] = dummy[dim+1];
		delete []dummy;
		dim = pos;
		examples.set_dim(dim);
		i=4;
	      }
	      else{
		// read dense data line
		for(pos=0;pos<dim;pos++){
		  readnext(data_stream,s,delimiter);
		  if(s[0] == '\0'){
		    throw read_exception("Not enough attributes - could not read examples");
		  };
		  try{
		    new_example[pos]= string2svmfloat(s);
		  }
		  catch(...){
		    throw read_exception("Attribute is no number - could not read example");
		  }; 
		};
	      };
	    }
	    else if(i == where_y){
	      // read classification
	      readnext(data_stream,s,delimiter);
	      if(s[0] == '\0'){
		throw read_exception("Not enough attributes - could not read examples");
	      };
	      try{
		new_example[dim] = string2svmfloat(s);
	      }
	      catch(...){
		throw read_exception("Class is no number - could not read example");
	      };
	      examples.set_initialised_y();
	    }
	    else if(i == where_alpha){
	      // read alpha
	      readnext(data_stream,s,delimiter);
	      if(s[0] == '\0'){
		throw read_exception("Not enough attributes - could not read examples");
	      };
	      try{
		new_example[dim+1] = string2svmfloat(s);
	      }
	      catch(...){
		throw read_exception("Alpha is no number - could not read example");
	      };
	      examples.set_initialised_alpha();
	    };
	  };
	};
	// insert examples, if ok.
	if(pos==dim){
	  // example ok, insert (more checks here if reading of x is reworked!!!
	  examples.put_example(new_example);
	  count++;
	};
      }
      else{
	// line contains parameters
	data_stream >> s;	
	if((0 == strcmp("dimension",s)) || (0==strcmp("dim",s))){
	  // dimension already set => error
	  SVMINT new_dim;
	  data_stream >> new_dim;
	  examples.set_dim(new_dim);
	  dim = new_dim;
	  if(new_example != 0){ delete []new_example; };
	  new_example = new SVMFLOAT[dim+2];
	}
	else if(0 == strcmp("number",s)){
	  // number of examples, check later for consistency
	  data_stream >> given_total;
	  if(given_total > 0){
	    // (examples.the_set).reserve((examples.the_set).size() + given_total);
	    examples.resize(examples.size()+given_total);
	  };
	}
	else if(0==strcmp("b",s)){
	  // hyperplane constant
	  data_stream >> s;
	  examples.b = string2svmfloat(s);
	}
	else if(0==strcmp("delimiter",s)){
	  data_stream >> s;
	  if((s[0] != '\0') && (s[1] != '\0')){
	    delimiter = s[1];
	  }
	  else if ((s[1] == '\0') && (s[0] != '\0')){
	    delimiter = s[0];
	    if(' ' == data_stream.peek()){
	      // if delimiter = ' ' we have only read one '
	      data_stream.get();
	      if(delimiter == data_stream.peek()){
		data_stream.get();
		delimiter = ' ';
	      };
	    };
	  }
	  else{
	    delimiter = ' ';
	  };
	  examples.my_format.delimiter = delimiter;
	}
	else if(0==strcmp("format",s)){
	  data_stream >> s;
	  if(0==strcmp("sparse",s)){
	    sparse = 1;
	  }
	  else{
	    sparse = 0;
	    where_x = 0;
	    where_y = 0;
	    where_alpha = 0;
	    for(int i=0;s[i] != '\0';i++){
	      if('x' == s[i]){
		where_x = i+1;
	      }
	      else if('y' == s[i]){
		where_y = i+1;
	      }
	      else if('a' == s[i]){
		where_alpha = i+1;
	      }
	      else{
		throw read_exception("Invalid format for examples");
	      };
	    };
	    if(0 == where_x){
	      throw read_exception("Invalid format for examples: x must be given");
	    };
	  };
	  if(0 == where_y){ examples.has_y = 0; };
	  if(0 == where_alpha){ examples.has_alphas = 0; };
	  examples.my_format.sparse = sparse;
	  examples.my_format.where_x = where_x;
	  examples.my_format.where_y = where_y;
	  examples.my_format.where_alpha = where_alpha;
	}
	else{
	  char* t = new char[MAXCHAR];
	  strcpy(t,"Unknown parameter: ");
	  strcat(t,s);
	  throw read_exception(t);
	};
      };
    }
    catch(general_exception g){
      // re-throw own exceptions
      if(new_example) delete []new_example;
      throw g;
    }
    catch(...){
      if(new_example) delete []new_example;
      throw read_exception("Error while reading from stream");
    };
  };
  if(new_example) delete []new_example;
  examples.compress();

  // check for consistency
  if((0 < given_total) && (count != given_total)){
    cout<<"WARNING: Wrong number of examples read ("<<count<<" read instead of "<<given_total<<")."<<endl;
  };
  delete []s;
  delete []s2;

  return data_stream;
};



ostream& operator<< (ostream& data_stream, example_set_c& examples){
  // output examples
  data_stream << "# svm example set" << endl;
  data_stream << "dimension "<< examples.dim << endl;
  data_stream << "number "<< examples.examples_total << endl;
  data_stream << "b " << examples.b << endl;
  char delimiter = examples.my_format.delimiter;
  if(delimiter != ' '){
    data_stream<<"delimiter '"<<delimiter<<"'"<<endl;
  };
  SVMINT total = examples.examples_total; 
  SVMINT dim = examples.dim;
  SVMINT j=0;
  svm_example the_example;
  // output examples;
  if(examples.my_format.sparse){
    data_stream<<"format "<<examples.my_format<<endl;
    for(SVMINT i=0;i<total;i++){
      // output example i
      the_example = examples.get_example(i);
      if((examples.Exp != 0) && (examples.Var != 0)){
	for(SVMINT pos=0;pos<the_example.length-1;pos++){
	  // output x_j
	  j = the_example.example[pos].index;
	  data_stream<<(the_example.example[pos].index+1)<<":";
	  if(0 != examples.Var[j]){
	    data_stream<<(the_example.example[pos].att*examples.Var[j]+examples.Exp[j]);
	  }
	  else{
	    data_stream<<the_example.example[pos].att+examples.Exp[j];
	  };
	  data_stream<<delimiter;
	}
	data_stream<<(the_example.example[the_example.length-1].index+1)<<":";
	  if(0 != examples.Var[dim-1]){
	    data_stream<<(the_example.example[the_example.length-1].att*examples.Var[dim-1]+examples.Exp[dim-1]);
	  }
	  else{
	    data_stream<<the_example.example[the_example.length-1].att+examples.Exp[dim-1];
	  };
      }
      else{
	for(SVMINT pos=0;pos<the_example.length-1;pos++){
	  data_stream<<(the_example.example[pos].index+1)<<":"
		     <<(the_example.example[pos].att)<<delimiter;
	};
	data_stream<<(the_example.example[the_example.length-1].index+1)<<":"
		   <<(the_example.example[the_example.length-1].att);
      };
      if(examples.has_y){
	data_stream << delimiter << "y:" << examples.get_y(i);
      };
      if(examples.has_alphas){
	if(examples.get_alpha(i) != 0){
	  data_stream << delimiter << "a:" << examples.get_alpha(i);
	};
      };
      data_stream << endl;
    };
  }
  else{
    // output dense format
    int where_x = examples.my_format.where_x;
    int where_y = examples.my_format.where_y;
    int where_alpha = examples.my_format.where_alpha;

    // output computed values as well
    if((0 == where_y) && (examples.initialised_y())){
      examples.my_format.where_y = 4;
      where_y = 4;
    }
    if((0 == where_alpha) && (examples.initialised_alpha())){
      examples.my_format.where_alpha = 5;
      where_alpha = 5;
    }
    data_stream<<"format "<<examples.my_format<<endl;
    SVMINT pos;
    for(SVMINT i=0;i<total;i++){
      // output example i      
      the_example = examples.get_example(i);
      for(int s=1;s<=5;s++){
	if(where_x == s){
	  if(1 != s) data_stream<<delimiter;
	  pos=0; // index in example (0..the_example.length-1
	  for(j=0;j<dim;j++){
	    // output attribute j
	    if(j != 0) data_stream<<delimiter;
	    if(the_example.example[pos].index == j){
	      // output the_example.example[pos].att
	      if((examples.Exp != 0) && (examples.Var != 0)){
		if(0 != examples.Var[j]){
		  data_stream<<(the_example.example[pos].att*examples.Var[j]+examples.Exp[j]);
		}
		else{
		  data_stream<<the_example.example[pos].att+examples.Exp[j];
		};
	      }
	      else{
		data_stream<<the_example.example[pos].att;
	      };
	      if(pos<the_example.length-1) pos++;
	    }
	    else{
	      data_stream<<"0";
	    };
	  };
	}
	else if(where_y == s){
	  if(1 != s) data_stream<<delimiter;
	  data_stream<<examples.get_y(i);
	}
	else if (where_alpha == s){
	  if(1 != s) data_stream<<delimiter;
	  data_stream<<examples.get_alpha(i);
	};
      };
      data_stream<<endl;
    };
  };
  return data_stream;
};