Geek Answers Handbook

Liblinear (in java) a simple example will not work

I am trying to use the liblinear library (java), and I am using a super-simple example with the template found here . An example of a case is to determine if the shape is a square or a rectangle.

here is my code:

import java.io.File; import java.io.IOException; import de.bwaldvogel.liblinear.Feature; import de.bwaldvogel.liblinear.FeatureNode; import de.bwaldvogel.liblinear.Linear; import de.bwaldvogel.liblinear.Model; import de.bwaldvogel.liblinear.Parameter; import de.bwaldvogel.liblinear.Problem; import de.bwaldvogel.liblinear.SolverType; public class Main { static int NUM_OF_TS_EXAMPLES = 8; // 1 = square, -1 = non-square static double[] GROUPS_ARRAY = {1, 1, 1, 1, -1, -1, -1, -1}; // squares static FeatureNode[] shape1 = {new FeatureNode(1, 2), new FeatureNode(2, 2)}; static FeatureNode[] shape2 = {new FeatureNode(1, 4), new FeatureNode(2, 4)}; static FeatureNode[] shape3 = {new FeatureNode(1, 9), new FeatureNode(2, 9)}; static FeatureNode[] shape4 = {new FeatureNode(1, 10), new FeatureNode(2, 10)}; // not squares static FeatureNode[] shape5 = {new FeatureNode(1, 5), new FeatureNode(2, 6)}; static FeatureNode[] shape6 = {new FeatureNode(1, 3), new FeatureNode(2, 4)}; static FeatureNode[] shape7 = {new FeatureNode(1, 6), new FeatureNode(2, 9)}; static FeatureNode[] shape8 = {new FeatureNode(1, 4), new FeatureNode(2, 2)}; // unknown squares static FeatureNode[] unkown1 = {new FeatureNode(1, 32), new FeatureNode(2, 32)}; static FeatureNode[] unkown2 = {new FeatureNode(1, 4), new FeatureNode(2, 2)}; static FeatureNode[] unkown3 = {new FeatureNode(1, 4), new FeatureNode(2, 2)}; static FeatureNode[][] trainingSetWithUnknown = { shape1, shape2, shape3, shape4, shape5, shape6, shape7, shape8 }; public static void main(String[] args) throws IOException { Problem problem = new Problem(); // number of training examples problem.l = NUM_OF_TS_EXAMPLES; // number of features problem.n = NUM_OF_TS_EXAMPLES + 1; // problem.x = ... // feature nodes problem.x = trainingSetWithUnknown; // problem.y = ... // target values problem.y = GROUPS_ARRAY; SolverType solver = SolverType.L2R_LR; // -s 0 double C = 1.0; // cost of constraints violation double eps = 0.01; // stopping criteria Parameter parameter = new Parameter(solver, C, eps); Model model = Linear.train(problem, parameter); File modelFile = new File("model"); model.save(modelFile); // load model or use it directly model = Model.load(modelFile); Feature[] instance = new FeatureNode[5]; double prediction = Linear.predict(model, instance); } }

After starting, I get this result in my console:

 iter 1 act 1.969e-02 pre 1.966e-02 delta 4.283e-03 f 5.545e+00 |g| 9.192e+00 CG 1 Exception in thread "main" java.lang.NullPointerException at de.bwaldvogel.liblinear.Linear.predictValues(Linear.java:370) at de.bwaldvogel.liblinear.Linear.predict(Linear.java:316) at Main.main(Main.java:73)

What am I doing wrong?

+4
source share
2 answers

I assume that the reason for the NullPointerException is that you are passing an uninitialized array of functions to the prediction function. First try to initialize instance :

 Feature[] instance = { new FeatureNode(1, 4), new FeatureNode(2, 2) };
+1
source

besides the fact that the NullPointer problem, the example leads only to strange predictions, mainly due to the wrong number of functions, here is my version:

 public class LibLinearTest { // 1 = quadratic, -1 = non-quadratic static double[] GROUPS_ARRAY = {1, 1, 1, 1, -1, -1, -1, -1}; // quadratic static FeatureNode[] tp1 = {new FeatureNode(1, 2), new FeatureNode(2, 4)}; static FeatureNode[] tp2 = {new FeatureNode(1, 4), new FeatureNode(2, 8)}; static FeatureNode[] tp3 = {new FeatureNode(1, 9), new FeatureNode(2, 81)}; static FeatureNode[] tp4 = {new FeatureNode(1, 10), new FeatureNode(2, 100)}; // not quadratic static FeatureNode[] tp5 = {new FeatureNode(1, 5), new FeatureNode(2, 6)}; static FeatureNode[] tp6 = {new FeatureNode(1, 3), new FeatureNode(2, 4)}; static FeatureNode[] tp7 = {new FeatureNode(1, 6), new FeatureNode(2, 9)}; static FeatureNode[] tp8 = {new FeatureNode(1, 4), new FeatureNode(2, 2)}; // unknown static FeatureNode[] up1 = {new FeatureNode(1, 32), new FeatureNode(2, 32)}; static FeatureNode[] up2 = {new FeatureNode(1, 5), new FeatureNode(2, 25)}; static FeatureNode[] up3 = {new FeatureNode(1, 4), new FeatureNode(2, 2)}; static FeatureNode[][] trainingSetWithUnknown = { tp1, tp2, tp3, tp4, tp5, tp6, tp7, tp8 }; public static void main(String[] args) throws IOException { Problem problem = new Problem(); problem.l = trainingSetWithUnknown.length; problem.n = 2; problem.x = trainingSetWithUnknown; problem.y = GROUPS_ARRAY; SolverType solver = SolverType.L2R_LR; // -s 0 double C = 1.0; // cost of constraints violation double eps = 0.001; // stopping criteria Parameter parameter = new Parameter(solver, C, eps); Model model = Linear.train(problem, parameter); Feature[] instance = tp1; double prediction = Linear.predict(model, instance); System.out.println("prediction : " + prediction); } }
0
source

More articles:


All Articles