File size: 7,579 Bytes
6f3ebfa | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 | package org.maltparser.ml.lib;
import java.io.Serializable;
import libsvm.svm_model;
import libsvm.svm_node;
import libsvm.svm_parameter;
import libsvm.svm_problem;
/**
* <p>This class borrows code from libsvm.svm.java of the Java implementation of the libsvm package.
* MaltLibsvmModel stores the model obtained from the training procedure. In addition to the original code the model is more integrated to
* MaltParser. Instead of moving features from MaltParser's internal data structures to liblinear's data structure it uses MaltParser's data
* structure directly on the model. </p>
*
* @author Johan Hall
*
*/
public class MaltLibsvmModel implements Serializable, MaltLibModel {
private static final long serialVersionUID = 7526471155622776147L;
public svm_parameter param; // parameter
public int nr_class; // number of classes, = 2 in regression/one class svm
public int l; // total #SV
public svm_node[][] SV; // SVs (SV[l])
public double[][] sv_coef; // coefficients for SVs in decision functions (sv_coef[k-1][l])
public double[] rho; // constants in decision functions (rho[k*(k-1)/2])
// for classification only
public int[] label; // label of each class (label[k])
public int[] nSV; // number of SVs for each class (nSV[k])
// nSV[0] + nSV[1] + ... + nSV[k-1] = l
public int[] start;
public MaltLibsvmModel(svm_model model, svm_problem problem) {
this.param = model.param;
this.nr_class = model.nr_class;
this.l = model.l;
this.SV = model.SV;
this.sv_coef = model.sv_coef;
this.rho = model.rho;
this.label = model.label;
this.nSV = model.nSV;
start = new int[nr_class];
start[0] = 0;
for(int i=1;i<nr_class;i++) {
start[i] = start[i-1]+nSV[i-1];
}
}
public int[] predict(MaltFeatureNode[] x) {
final double[] dec_values = new double[nr_class*(nr_class-1)/2];
final double[] kvalue = new double[l];
final int[] vote = new int[nr_class];
int i;
for(i=0;i<l;i++) {
kvalue[i] = MaltLibsvmModel.k_function(x,SV[i],param);
}
for(i=0;i<nr_class;i++) {
vote[i] = 0;
}
int p=0;
for(i=0;i<nr_class;i++) {
for(int j=i+1;j<nr_class;j++) {
double sum = 0;
int si = start[i];
int sj = start[j];
int ci = nSV[i];
int cj = nSV[j];
int k;
double[] coef1 = sv_coef[j-1];
double[] coef2 = sv_coef[i];
for(k=0;k<ci;k++)
sum += coef1[si+k] * kvalue[si+k];
for(k=0;k<cj;k++)
sum += coef2[sj+k] * kvalue[sj+k];
sum -= rho[p];
dec_values[p] = sum;
if(dec_values[p] > 0)
++vote[i];
else
++vote[j];
p++;
}
}
final int[] predictionList = new int[nr_class];
System.arraycopy(label, 0, predictionList, 0, nr_class);
int tmp;
int iMax;
final int nc = nr_class-1;
for (i=0; i < nc; i++) {
iMax = i;
for (int j=i+1; j < nr_class; j++) {
if (vote[j] > vote[iMax]) {
iMax = j;
}
}
if (iMax != i) {
tmp = vote[iMax];
vote[iMax] = vote[i];
vote[i] = tmp;
tmp = predictionList[iMax];
predictionList[iMax] = predictionList[i];
predictionList[i] = tmp;
}
}
return predictionList;
}
public int predict_one(MaltFeatureNode[] x) {
final double[] dec_values = new double[nr_class*(nr_class-1)/2];
final double[] kvalue = new double[l];
final int[] vote = new int[nr_class];
int i;
for(i=0;i<l;i++) {
kvalue[i] = MaltLibsvmModel.k_function(x,SV[i],param);
}
for(i=0;i<nr_class;i++) {
vote[i] = 0;
}
int p=0;
for(i=0;i<nr_class;i++) {
for(int j=i+1;j<nr_class;j++) {
double sum = 0;
int si = start[i];
int sj = start[j];
int ci = nSV[i];
int cj = nSV[j];
int k;
double[] coef1 = sv_coef[j-1];
double[] coef2 = sv_coef[i];
for(k=0;k<ci;k++)
sum += coef1[si+k] * kvalue[si+k];
for(k=0;k<cj;k++)
sum += coef2[sj+k] * kvalue[sj+k];
sum -= rho[p];
dec_values[p] = sum;
if(dec_values[p] > 0)
++vote[i];
else
++vote[j];
p++;
}
}
int max = vote[0];
int max_index = 0;
for (i = 1; i < vote.length; i++) {
if (vote[i] > max) {
max = vote[i];
max_index = i;
}
}
return label[max_index];
}
static double dot(MaltFeatureNode[] x, svm_node[] y) {
double sum = 0;
final int xlen = x.length;
final int ylen = y.length;
int i = 0;
int j = 0;
while(i < xlen && j < ylen)
{
if(x[i].index == y[j].index)
sum += x[i++].value * y[j++].value;
else
{
if(x[i].index > y[j].index)
++j;
else
++i;
}
}
return sum;
}
static double powi(double base, int times) {
double tmp = base, ret = 1.0;
for(int t=times; t>0; t/=2)
{
if(t%2==1) ret*=tmp;
tmp = tmp * tmp;
}
return ret;
}
static double k_function(MaltFeatureNode[] x, svm_node[] y, svm_parameter param) {
switch(param.kernel_type)
{
case svm_parameter.LINEAR:
return dot(x,y);
case svm_parameter.POLY:
return powi(param.gamma*dot(x,y)+param.coef0,param.degree);
case svm_parameter.RBF:
{
double sum = 0;
int xlen = x.length;
int ylen = y.length;
int i = 0;
int j = 0;
while(i < xlen && j < ylen)
{
if(x[i].index == y[j].index)
{
double d = x[i++].value - y[j++].value;
sum += d*d;
}
else if(x[i].index > y[j].index)
{
sum += y[j].value * y[j].value;
++j;
}
else
{
sum += x[i].value * x[i].value;
++i;
}
}
while(i < xlen)
{
sum += x[i].value * x[i].value;
++i;
}
while(j < ylen)
{
sum += y[j].value * y[j].value;
++j;
}
return Math.exp(-param.gamma*sum);
}
case svm_parameter.SIGMOID:
return Math.tanh(param.gamma*dot(x,y)+param.coef0);
case svm_parameter.PRECOMPUTED:
return x[(int)(y[0].value)].value;
default:
return 0; // java
}
}
public int[] getLabels() {
if (label != null) {
final int[] labels = new int[nr_class];
for(int i=0;i<nr_class;i++) {
labels[i] = label[i];
}
return labels;
}
return null;
}
}
|