liblinear-ruby 0.0.4 → 0.0.5

data/{liblinear-1.94 → liblinear-1.95}/blas/blasp.h

@@ -3,6 +3,10 @@
 
 /* Functions  listed in alphabetical order */
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 #ifdef F2C_COMPAT
 
 void cdotc_(fcomplex *dotval, int *n, fcomplex *cx, int *incx,
@@ -428,3 +432,7 @@ int ztrsm_(char *side, char *uplo, char *transa, char *diag, int *m,
 
 int ztrsv_(char *uplo, char *trans, char *diag, int *n, dcomplex *a,
            int *lda, dcomplex *x, int *incx);
+
+#ifdef __cplusplus
+}
+#endif

data/{liblinear-1.94 → liblinear-1.95}/blas/daxpy.c

@@ -1,5 +1,9 @@
 #include "blas.h"
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 int daxpy_(int *n, double *sa, double *sx, int *incx, double *sy,
            int *incy)
 {
@@ -47,3 +51,7 @@ int daxpy_(int *n, double *sa, double *sx, int *incx, double *sy,
 
   return 0;
 } /* daxpy_ */
+
+#ifdef __cplusplus
+}
+#endif

data/{liblinear-1.94 → liblinear-1.95}/blas/ddot.c

@@ -1,5 +1,9 @@
 #include "blas.h"
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 double ddot_(int *n, double *sx, int *incx, double *sy, int *incy)
 {
   long int i, m, nn, iincx, iincy;
@@ -48,3 +52,7 @@ double ddot_(int *n, double *sx, int *incx, double *sy, int *incy)
 
   return stemp;
 } /* ddot_ */
+
+#ifdef __cplusplus
+}
+#endif

data/{liblinear-1.94 → liblinear-1.95}/blas/dnrm2.c

@@ -1,6 +1,10 @@
 #include <math.h>  /* Needed for fabs() and sqrt() */
 #include "blas.h"
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 double dnrm2_(int *n, double *x, int *incx)
 {
   long int ix, nn, iincx;
@@ -60,3 +64,7 @@ double dnrm2_(int *n, double *x, int *incx)
   return norm;
 
 } /* dnrm2_ */
+
+#ifdef __cplusplus
+}
+#endif

data/{liblinear-1.94 → liblinear-1.95}/blas/dscal.c

@@ -1,5 +1,9 @@
 #include "blas.h"
 
+#ifdef __cplusplus
+extern "C" {
+#endif
+
 int dscal_(int *n, double *sa, double *sx, int *incx)
 {
   long int i, m, nincx, nn, iincx;
@@ -42,3 +46,7 @@ int dscal_(int *n, double *sa, double *sx, int *incx)
 
   return 0;
 } /* dscal_ */
+
+#ifdef __cplusplus
+}
+#endif

data/{liblinear-1.94 → liblinear-1.95}/linear.cpp

@@ -1010,7 +1010,7 @@ static void solve_l2r_l1l2_svr(
 	double d, G, H;
 	double Gmax_old = INF;
 	double Gmax_new, Gnorm1_new;
-	double Gnorm1_init;
+	double Gnorm1_init = -1.0; // Gnorm1_init is initialized at the first iteration
 	double *beta = new double[l];
 	double *QD = new double[l];
 	double *y = prob->y;
@@ -1409,7 +1409,7 @@ static void solve_l1r_l2_svc(
 	double d, G_loss, G, H;
 	double Gmax_old = INF;
 	double Gmax_new, Gnorm1_new;
-	double Gnorm1_init;
+	double Gnorm1_init = -1.0; // Gnorm1_init is initialized at the first iteration
 	double d_old, d_diff;
 	double loss_old, loss_new;
 	double appxcond, cond;
@@ -1699,7 +1699,7 @@ static void solve_l1r_lr(
 	double sigma = 0.01;
 	double w_norm, w_norm_new;
 	double z, G, H;
-	double Gnorm1_init;
+	double Gnorm1_init = -1.0; // Gnorm1_init is initialized at the first iteration
 	double Gmax_old = INF;
 	double Gmax_new, Gnorm1_new;
 	double QP_Gmax_old = INF;
@@ -2051,8 +2051,8 @@ static void transpose(const problem *prob, feature_node **x_space_ret, problem *
 	int i;
 	int l = prob->l;
 	int n = prob->n;
-	long int nnz = 0;
-	long int *col_ptr = new long int [n+1];
+	size_t nnz = 0;
+	size_t *col_ptr = new size_t [n+1];
 	feature_node *x_space;
 	prob_col->l = l;
 	prob_col->n = n;
@@ -2305,9 +2305,7 @@ model* train(const problem *prob, const parameter *param)
 	model_->param = *param;
 	model_->bias = prob->bias;
 
-	if(param->solver_type == L2R_L2LOSS_SVR ||
-	   param->solver_type == L2R_L1LOSS_SVR_DUAL ||
-	   param->solver_type == L2R_L2LOSS_SVR_DUAL)
+	if(check_regression_model(model_))
 	{
 		model_->w = Malloc(double, w_size);
 		model_->nr_class = 2;
@@ -2512,9 +2510,7 @@ double predict_values(const struct model *model_, const struct feature_node *x,
 
 	if(nr_class==2)
 	{
-		if(model_->param.solver_type == L2R_L2LOSS_SVR ||
-		   model_->param.solver_type == L2R_L1LOSS_SVR_DUAL ||
-		   model_->param.solver_type == L2R_L2LOSS_SVR_DUAL)
+		if(check_regression_model(model_))
 			return dec_values[0];
 		else
 			return (dec_values[0]>0)?model_->label[0]:model_->label[1];
@@ -2764,6 +2760,53 @@ void get_labels(const model *model_, int* label)
 			label[i] = model_->label[i];
 }
 
+// use inline here for better performance (around 20% faster than the non-inline one)
+static inline double get_w_value(const struct model *model_, int idx, int label_idx) 
+{
+	int nr_class = model_->nr_class;
+	int solver_type = model_->param.solver_type;
+	const double *w = model_->w;
+
+	if(idx < 0 || idx > model_->nr_feature)
+		return 0;
+	if(check_regression_model(model_))
+		return w[idx];
+	else 
+	{
+		if(label_idx < 0 || label_idx >= nr_class)
+			return 0;
+		if(nr_class == 2 && solver_type != MCSVM_CS)
+		{
+			if(label_idx == 0)
+				return w[idx];
+			else
+				return -w[idx];
+		}
+		else
+			return w[idx*nr_class+label_idx];
+	}
+}
+
+// feat_idx: starting from 1 to nr_feature
+// label_idx: starting from 0 to nr_class-1 for classification models;
+//            for regression models, label_idx is ignored.
+double get_decfun_coef(const struct model *model_, int feat_idx, int label_idx)
+{
+	if(feat_idx > model_->nr_feature)
+		return 0;
+	return get_w_value(model_, feat_idx-1, label_idx);
+}
+
+double get_decfun_bias(const struct model *model_, int label_idx)
+{
+	int bias_idx = model_->nr_feature;
+	double bias = model_->bias;
+	if(bias <= 0)
+		return 0;
+	else
+		return bias*get_w_value(model_, bias_idx, label_idx);
+}
+
 void free_model_content(struct model *model_ptr)
 {
 	if(model_ptr->w != NULL)
@@ -2824,6 +2867,13 @@ int check_probability_model(const struct model *model_)
 			model_->param.solver_type==L1R_LR);
 }
 
+int check_regression_model(const struct model *model_)
+{
+	return (model_->param.solver_type==L2R_L2LOSS_SVR ||
+			model_->param.solver_type==L2R_L1LOSS_SVR_DUAL ||
+			model_->param.solver_type==L2R_L2LOSS_SVR_DUAL);
+}
+
 void set_print_string_function(void (*print_func)(const char*))
 {
 	if (print_func == NULL)

data/{liblinear-1.94 → liblinear-1.95}/linear.def

@@ -16,3 +16,6 @@ EXPORTS
 	check_parameter	@14
 	check_probability_model	@15
 	set_print_string_function	@16
+    get_decfun_coef @17
+    get_decfun_bias @18
+    check_regression_model  @19

data/{liblinear-1.94 → liblinear-1.95}/linear.h

@@ -57,6 +57,8 @@ struct model *load_model(const char *model_file_name);
 int get_nr_feature(const struct model *model_);
 int get_nr_class(const struct model *model_);
 void get_labels(const struct model *model_, int* label);
+double get_decfun_coef(const struct model *model_, int feat_idx, int label_idx);
+double get_decfun_bias(const struct model *model_, int label_idx);
 
 void free_model_content(struct model *model_ptr);
 void free_and_destroy_model(struct model **model_ptr_ptr);
@@ -64,6 +66,7 @@ void destroy_param(struct parameter *param);
 
 const char *check_parameter(const struct problem *prob, const struct parameter *param);
 int check_probability_model(const struct model *model);
+int check_regression_model(const struct model *model);
 void set_print_string_function(void (*print_func) (const char*));
 
 #ifdef __cplusplus

data/{liblinear-1.94 → liblinear-1.95}/matlab/Makefile

@@ -7,26 +7,17 @@ CC ?= gcc
 CFLAGS = -Wall -Wconversion -O3 -fPIC -I$(MATLABDIR)/extern/include -I..
 
 MEX = $(MATLABDIR)/bin/mex
-MEX_OPTION = CC\#$(CXX) CXX\#$(CXX) CFLAGS\#"$(CFLAGS)" CXXFLAGS\#"$(CFLAGS)"
+MEX_OPTION = CC="$(CXX)" CXX="$(CXX)" CFLAGS="$(CFLAGS)" CXXFLAGS="$(CFLAGS)"
 # comment the following line if you use MATLAB on a 32-bit computer
 MEX_OPTION += -largeArrayDims
 MEX_EXT = $(shell $(MATLABDIR)/bin/mexext)
 
-OCTAVEDIR ?= /usr/include/octave
-OCTAVE_MEX = env CC=$(CXX) mkoctfile
-OCTAVE_MEX_OPTION = --mex
-OCTAVE_MEX_EXT = mex
-OCTAVE_CFLAGS = -Wall -O3 -fPIC -I$(OCTAVEDIR) -I..
-
 all:	matlab
 
 matlab:	binary
 
 octave:
-	@make MEX="$(OCTAVE_MEX)" MEX_OPTION="$(OCTAVE_MEX_OPTION)" \
-	MEX_EXT="$(OCTAVE_MEX_EXT)" CFLAGS="$(OCTAVE_CFLAGS)" \
-	binary
-
+	@echo "please type make under Octave"
 binary: train.$(MEX_EXT) predict.$(MEX_EXT) libsvmread.$(MEX_EXT) libsvmwrite.$(MEX_EXT)
 
 train.$(MEX_EXT): train.c ../linear.h ../tron.o ../linear.o linear_model_matlab.o ../blas/blas.a

data/{liblinear-1.94 → liblinear-1.95}/matlab/libsvmread.c

@@ -56,14 +56,13 @@ static char* readline(FILE *input)
 // read in a problem (in libsvm format)
 void read_problem(const char *filename, int nlhs, mxArray *plhs[])
 {
-	int max_index, min_index, inst_max_index, i;
-	long elements, k;
+	int max_index, min_index, inst_max_index;
+	size_t elements, k, i, l=0;
 	FILE *fp = fopen(filename,"r");
-	int l = 0;
 	char *endptr;
 	mwIndex *ir, *jc;
 	double *labels, *samples;
-	
+
 	if(fp == NULL)
 	{
 		mexPrintf("can't open input file %s\n",filename);

data/{liblinear-1.94 → liblinear-1.95}/matlab/libsvmwrite.c

@@ -26,9 +26,8 @@ static void fake_answer(int nlhs, mxArray *plhs[])
 void libsvmwrite(const char *filename, const mxArray *label_vec, const mxArray *instance_mat)
 {
 	FILE *fp = fopen(filename,"w");
-	int i, k, low, high, l;
-	mwIndex *ir, *jc;
-	int label_vector_row_num;
+	mwIndex *ir, *jc, k, low, high;
+	size_t i, l, label_vector_row_num;
 	double *samples, *labels;
 	mxArray *instance_mat_col; // instance sparse matrix in column format
 
@@ -52,8 +51,8 @@ void libsvmwrite(const char *filename, const mxArray *label_vec, const mxArray *
 	}
 
 	// the number of instance
-	l = (int) mxGetN(instance_mat_col);
-	label_vector_row_num = (int) mxGetM(label_vec);
+	l = mxGetN(instance_mat_col);
+	label_vector_row_num = mxGetM(label_vec);
 
 	if(label_vector_row_num!=l)
 	{
@@ -71,9 +70,9 @@ void libsvmwrite(const char *filename, const mxArray *label_vec, const mxArray *
 	{
 		fprintf(fp,"%g", labels[i]);
 
-		low = (int) jc[i], high = (int) jc[i+1];
+		low = jc[i], high = jc[i+1];
 		for(k=low;k<high;k++)
-			fprintf(fp," %ld:%g", ir[k]+1, samples[k]);		
+			fprintf(fp," %zu:%g", (size_t)ir[k]+1, samples[k]);		
 
 		fprintf(fp,"\n");
 	}

data/{liblinear-1.94 → liblinear-1.95}/matlab/linear_model_matlab.c

@@ -145,7 +145,7 @@ const char *matlab_matrix_to_model(struct model *model_, const mxArray *matlab_s
 
 	// bias
 	ptr = mxGetPr(rhs[id]);
-	model_->bias = (int)ptr[0];
+	model_->bias = ptr[0];
 	id++;
 
 	if(model_->bias>=0)

data/{liblinear-1.94 → liblinear-1.95}/matlab/make.m

@@ -1,21 +1,21 @@
-% This make.m is for MATLAB and OCTAVE under Windows, Mac, and Unix
-
-try
-	Type = ver;
-	% This part is for OCTAVE
-	if(strcmp(Type(1).Name, 'Octave') == 1)
-		mex libsvmread.c
-		mex libsvmwrite.c
-		mex train.c linear_model_matlab.c ../linear.cpp ../tron.cpp ../blas/*.c
-		mex predict.c linear_model_matlab.c ../linear.cpp ../tron.cpp ../blas/*.c
-	% This part is for MATLAB
-	% Add -largeArrayDims on 64-bit machines of MATLAB
-	else
-		mex CFLAGS="\$CFLAGS -std=c99" -largeArrayDims libsvmread.c
-		mex CFLAGS="\$CFLAGS -std=c99" -largeArrayDims libsvmwrite.c
-		mex CFLAGS="\$CFLAGS -std=c99" -largeArrayDims train.c linear_model_matlab.c ../linear.cpp ../tron.cpp "../blas/*.c"
-		mex CFLAGS="\$CFLAGS -std=c99" -largeArrayDims predict.c linear_model_matlab.c ../linear.cpp ../tron.cpp "../blas/*.c"
-	end
-catch
-	fprintf('If make.m fails, please check README about detailed instructions.\n');
-end
+% This make.m is for MATLAB and OCTAVE under Windows, Mac, and Unix
+
+try
+	Type = ver;
+	% This part is for OCTAVE
+	if(strcmp(Type(1).Name, 'Octave') == 1)
+		mex libsvmread.c
+		mex libsvmwrite.c
+		mex train.c linear_model_matlab.c ../linear.cpp ../tron.cpp ../blas/daxpy.c ../blas/ddot.c ../blas/dnrm2.c ../blas/dscal.c
+		mex predict.c linear_model_matlab.c ../linear.cpp ../tron.cpp ../blas/daxpy.c ../blas/ddot.c ../blas/dnrm2.c ../blas/dscal.c
+	% This part is for MATLAB
+	% Add -largeArrayDims on 64-bit machines of MATLAB
+	else
+		mex CFLAGS="\$CFLAGS -std=c99" -largeArrayDims libsvmread.c
+		mex CFLAGS="\$CFLAGS -std=c99" -largeArrayDims libsvmwrite.c
+		mex CFLAGS="\$CFLAGS -std=c99" -largeArrayDims train.c linear_model_matlab.c ../linear.cpp ../tron.cpp ../blas/daxpy.c ../blas/ddot.c ../blas/dnrm2.c ../blas/dscal.c
+		mex CFLAGS="\$CFLAGS -std=c99" -largeArrayDims predict.c linear_model_matlab.c ../linear.cpp ../tron.cpp ../blas/daxpy.c ../blas/ddot.c ../blas/dnrm2.c ../blas/dscal.c
+	end
+catch
+	fprintf('If make.m fails, please check README about detailed instructions.\n');
+end

data/{liblinear-1.94 → liblinear-1.95}/matlab/predict.c

@@ -23,8 +23,8 @@ int col_format_flag;
 
 void read_sparse_instance(const mxArray *prhs, int index, struct feature_node *x, int feature_number, double bias)
 {
-	int i, j, low, high;
-	mwIndex *ir, *jc;
+	int j;
+	mwIndex *ir, *jc, low, high, i;
 	double *samples;
 
 	ir = mxGetIr(prhs);
@@ -33,7 +33,7 @@ void read_sparse_instance(const mxArray *prhs, int index, struct feature_node *x
 
 	// each column is one instance
 	j = 0;
-	low = (int) jc[index], high = (int) jc[index+1];
+	low = jc[index], high = jc[index+1];
 	for(i=low; i<high && (int) (ir[i])<feature_number; i++)
 	{
 		x[j].index = (int) ir[i]+1;
@@ -176,9 +176,7 @@ void do_predict(int nlhs, mxArray *plhs[], const mxArray *prhs[], struct model *
 		++total;
 	}
 
-	if(model_->param.solver_type==L2R_L2LOSS_SVR ||
-	   model_->param.solver_type==L2R_L1LOSS_SVR_DUAL ||
-	   model_->param.solver_type==L2R_L2LOSS_SVR_DUAL)
+	if(check_regression_model(model_))
 	{
 		info("Mean squared error = %g (regression)\n",error/total);
 		info("Squared correlation coefficient = %g (regression)\n",

data/{liblinear-1.94 → liblinear-1.95}/matlab/train.c

@@ -252,9 +252,10 @@ static void fake_answer(int nlhs, mxArray *plhs[])
 
 int read_problem_sparse(const mxArray *label_vec, const mxArray *instance_mat)
 {
-	int i, j, k, low, high;
-	mwIndex *ir, *jc;
-	int elements, max_index, num_samples, label_vector_row_num;
+	mwIndex *ir, *jc, low, high, k;
+	// using size_t due to the output type of matlab functions
+	size_t i, j, l, elements, max_index, label_vector_row_num;
+	mwSize num_samples;
 	double *samples, *labels;
 	mxArray *instance_mat_col; // instance sparse matrix in column format
 
@@ -279,10 +280,11 @@ int read_problem_sparse(const mxArray *label_vec, const mxArray *instance_mat)
 	}
 
 	// the number of instance
-	prob.l = (int) mxGetN(instance_mat_col);
-	label_vector_row_num = (int) mxGetM(label_vec);
+	l = mxGetN(instance_mat_col);
+	label_vector_row_num = mxGetM(label_vec);
+	prob.l = (int) l;
 
-	if(label_vector_row_num!=prob.l)
+	if(label_vector_row_num!=l)
 	{
 		mexPrintf("Length of label vector does not match # of instances.\n");
 		return -1;
@@ -294,23 +296,23 @@ int read_problem_sparse(const mxArray *label_vec, const mxArray *instance_mat)
 	ir = mxGetIr(instance_mat_col);
 	jc = mxGetJc(instance_mat_col);
 
-	num_samples = (int) mxGetNzmax(instance_mat_col);
+	num_samples = mxGetNzmax(instance_mat_col);
 
-	elements = num_samples + prob.l*2;
-	max_index = (int) mxGetM(instance_mat_col);
+	elements = num_samples + l*2;
+	max_index = mxGetM(instance_mat_col);
 
-	prob.y = Malloc(double, prob.l);
-	prob.x = Malloc(struct feature_node*, prob.l);
+	prob.y = Malloc(double, l);
+	prob.x = Malloc(struct feature_node*, l);
 	x_space = Malloc(struct feature_node, elements);
 
 	prob.bias=bias;
 
 	j = 0;
-	for(i=0;i<prob.l;i++)
+	for(i=0;i<l;i++)
 	{
 		prob.x[i] = &x_space[j];
 		prob.y[i] = labels[i];
-		low = (int) jc[i], high = (int) jc[i+1];
+		low = jc[i], high = jc[i+1];
 		for(k=low;k<high;k++)
 		{
 			x_space[j].index = (int) ir[k]+1;
@@ -319,7 +321,7 @@ int read_problem_sparse(const mxArray *label_vec, const mxArray *instance_mat)
 	 	}
 		if(prob.bias>=0)
 		{
-			x_space[j].index = max_index+1;
+			x_space[j].index = (int) max_index+1;
 			x_space[j].value = prob.bias;
 			j++;
 		}
@@ -327,9 +329,9 @@ int read_problem_sparse(const mxArray *label_vec, const mxArray *instance_mat)
 	}
 
 	if(prob.bias>=0)
-		prob.n = max_index+1;
+		prob.n = (int) max_index+1;
 	else
-		prob.n = max_index;
+		prob.n = (int) max_index;
 
 	return 0;
 }
@@ -356,12 +358,20 @@ void mexFunction( int nlhs, mxArray *plhs[],
 	{
 		int err=0;
 
-		if(!mxIsDouble(prhs[0]) || !mxIsDouble(prhs[1])) {
+		if(!mxIsDouble(prhs[0]) || !mxIsDouble(prhs[1]))
+		{
 			mexPrintf("Error: label vector and instance matrix must be double\n");
 			fake_answer(nlhs, plhs);
 			return;
 		}
 
+		if(mxIsSparse(prhs[0]))
+		{
+			mexPrintf("Error: label vector should not be in sparse format");
+			fake_answer(nlhs, plhs);
+			return;
+		}
+
 		if(parse_command_line(nrhs, prhs, NULL))
 		{
 			exit_with_help();