This repository is for Subspace Support Vector Data Description (S-SVDD). The codes are provided as .m (matlab) files to be executed in matlab. The codes are provided without any warranty or gurantee. Download the package from HERE, unzip and add the folder ssvdd-main to the path in matlab. see SSVDDdemo.m for exmaple usage.
S-SVDD requires LIBSVM for SVDD. Before executing the codes, make sure that correct version (3.22) of LIBSVM for SVDD is installed already. In order to install LIBSVM for SVDD Please download zip file from HERE, put sources into libsvm-3.22 available HERE, and make the code. For more details about how to install libsvm, please refer HERE
Possible inputs to ssvddtrain()
The first input argument is Training data
other options (input arguments) include
'maxIter' :Maximim iteraions, Default=100
'C' :Value of hyperparameter C, Default=0.1
'd' :data in lower dimension, make sure that input d<D, Default=1,
'eta' :Used as step size for gradient, Default=0.1
'psi' :regularization term, Default=1 i.e., No regularization term
:Other options for psi are 2,3,4 (Please refer to paper for more details)
'B' :Controling the importance of regularization term, Default=0.1
'npt' :1 for Non-linear Projection Trick (NPT)-based non-linear Subspace-SVDD (Default=0, linear)
's' :Hyperparameter for the kernel inside NPT (Default=0.001).
ssvddmodel=ssvddtrain(Traindata,'C',0.12,'d',2,'eta',0.02,'psi',3,'npt',1,'s',0.01);
[predicted_labels,accuracy,sensitivity,specificity]=ssvddtest(Testdata,testlabels,ssvddmodel);
If you use any part of this repository in your implementation, consider citing the following paper.
@inproceedings{sohrab2018subspace,
title={Subspace support vector data description},
author={Sohrab, Fahad and Raitoharju, Jenni and Gabbouj, Moncef and Iosifidis, Alexandros},
booktitle={2018 24th International Conference on Pattern Recognition (ICPR)},
pages={722--727},
year={2018},
organization={IEEE}
}
Please contact fahad.sohrab@tuni.fi for any issues.