search

boortel / SIFT-and-SURF-based-AD

Implementation of the "SIFT and SURF based feature extraction for the anomaly detection" paper
2 stars 0 forks source link

Why semi-svm 'nu' is 0? #3

Open Wangjyddddddd opened 1 year ago

Wangjyddddddd commented 1 year ago

%% One-class SVM

% OCC - SIFT [SVMModel1, score1, acc1_max, boundary1] = trainSVM(SIFT_trainOCC, labels_trainOCCNum, SIFT_test, labels_testNum, 0.04, 1);

% OCC - SURF [SVMModel2, score2, acc2_max, boundary2] = trainSVM(SURF_trainOCC, labels_trainOCCNum, SURF_test, labels_testNum, 0.04, 1);

% Semi-supervised - SIFT [SVMModel3, score3, acc3_max, boundary3] = trainSVM(SIFT_trainSemi, labels_trainSemiNum, SIFT_test, labels_testNum, 0, 0);

% Semi-supervised - SURF [SVMModel4, score4, acc4_max, boundary4] = trainSVM(SURF_trainSemi, labels_trainSemiNum, SURF_test, labels_testNum, 0, 0);

So why one-class svm 'nu' can gei 0.04, but semi-svm 'nu' is 0? Can I try other numbers in 0 to 1 of nu for semi?

boortel commented 1 year ago

Hello, nu = 0 by semi-supervised experiment might be some typo - feel free to experiment with other values as well. You can also try the more up-to-date Python implementation referred in the ReadMe, this one is not maintained any more.

You can also share your experimental results there and feel free to ask further.

Wangjyddddddd commented 1 year ago

Thanks a lot :)