LIBSVM中归一化和参数估计的重要性

LIBSVM中归一化和参数估计的重要性

LIBSVM中归一化和参数估计的重要性

下面是Libsvm guide中采用不同方法使用Libsvm的例子：（可以看到归一化和参数估计的重要性）

Astroparticle Physics

– Original sets with default parameters
$./svm-train train.1
$./svm-predict test.1 train.1.model test.1.predict
! Accuracy = 66.925%----很低

– Scaled sets with default parameters
$./svm-scale -l -1 -u 1 -s range1 train.1 > train.1.scale----归一化的结果
$./svm-scale -r range1 test.1 > test.1.scale
$./svm-train train.1.scale
$./svm-predict test.1.scale train.1.scale.model test.1.predict
! Accuracy = 96.15%--还不错

– Scaled sets with parameter selection
$python grid.py train.1.scale
· · ·
8192.0 0.0 96.9569

(Best C=8192.0, G=0.03125 with five-fold cross-validation rate=96.8922%)
$./svm-train -c 8192.0 -g 0.03125 train.1.scale----估计参数c和g
$./svm-predict test.1.scale train.1.scale.model test.1.predict
! Accuracy = 96.525%

– Using an automatic script
$python easy.py train.1 test.1
Scaling training data...
Cross validation...
Best c=8192.0, g=0.03125 

Training...
Scaling testing data...
Testing...

Accuracy = 96.525% (3875/4000) (classification)