本文不会介绍SVM的基本原理,如果想了解SVM基本原理,请参阅相关书籍。
要使用Java机器学习库Smile,需首先在项目的Maven配置文件pom.xml中添加如下的maven依赖项:
<dependency>
<groupId>com.github.haifengl</groupId>
<artifactId>smile-core</artifactId>
<version>1.4.0</version>
</dependency>
Smile 库的SVM类是一个泛型类型,默认情况下进行二分类,选择参数为核函数类型和惩罚项参数。
import smile.classification.SVM;
import smile.math.kernel.GaussianKernel;
public class Demo {
public static void main(String[]args){
double gamma = 1.0;
double C = 1.0;
//通过某种方式获取训练数据及其类标
double[][] data = ...
int[] label = ...
SVM<double[]> svm = new SVM<double[]>(
new GaussianKernel(gamma), C);
svm.learn(data, label); //训练模型
svm.finish();
//获取测试数据
double[][] testData = ...
int[] result = new int[testData.length];
for(int i=0; i < testData.length; i++){
result[i] = svm.predict(testData[i]);
}
}
}
接下来是我利用SVM对iris数据集进行分类的程序。首先我们将iris数据保存iris.txt文件,如下结构:
5.1 3.5 1.4 0.2 0
4.9 3 1.4 0.2 0
...
每一行代表一个测试数据项,前4列是属性向量,最后一列是类标(在Smile中类标不能为负数,并且只能是从0开始的正整数,所以上述类标为:0、1、2)。检测的完整的源代码如下:
import smile.classification.SVM;
import smile.math.kernel.GaussianKernel;
import java.io.BufferedReader;
import java.io.File;
import java.io.FileReader;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.List;
/**
* Created by zhanghuayan on 2017/1/16.
*/
public class ClassificationTest {
public static void main(String[] args) throws Exception {
List<List<Double>> datas =
new ArrayList<List<Double>>();
List<Double> data = new ArrayList<Double>();
List<Integer> labels = new ArrayList<Integer>();
String line;
List<String> lines;
File file = new File("iris.txt");
BufferedReader reader =
new BufferedReader(new FileReader(file));
while ((line = reader.readLine()) != null) {
lines = Arrays.asList(line.trim().split("\t"));
for (int i = 0; i < lines.size() - 1; i++) {
data.add(Double.parseDouble(lines.get(i)));
}
labels.add(Integer.parseInt(
lines.get(lines.size() - 1)));
datas.add(data);
data = new ArrayList<Double>();
}
//转换label
int[] label = new int[labels.size()];
for (int i = 0; i < label.length; i++) {
label[i] = labels.get(i);
}
//转换属性
int rows = datas.size();
int cols = datas.get(0).size();
double[][] srcData = new double[rows][cols];
for (int i = 0; i < rows; i++) {
for (int j = 0; j < cols; j++) {
srcData[i][j] = datas.get(i).get(j);
}
}
SVM<double[]> svm = new SVM<double[]>(
new GaussianKernel(1.0), 1.0, 3,
SVM.Multiclass.ONE_VS_ALL);
svm.learn(srcData, label);
svm.finish();
double right = 0;
for (int i = 0; i < srcData.length; i++) {
int tag = svm.predict(srcData[i]);
if (tag == label[i]) {
right += 1;
}
}
right = right / srcData.length;
System.out.println(
"Accrurate: " + right * 100 + "%");
}
}