Entertainment

ONE-CLASS DATASETS AND RESOURCES

AUTOMATIC COUNTER-EXAMPLE GENERATION-BASED

ONE-CLASS CLASSIFIER

The links and source code below are deprecated. Please contact me for more information about these. (banhalmi at inf.u-szeged.hu)

Automatic Counter-Example-Generation-Based Methods

One-class datasets

Protein classification benchmark collection: this benchmark is a collection of ranking problems containing very few positive examples in a high dimensional space.
Silent frames: This dataset is created by András Bánhalmi, and contains silent parts of speech samples. The feature vectors are Filter Bank transformed speech frames.
Unvoiced frames: This dataset is created by András Bánhalmi, and contains unvoiced parts of speech samples. The feature vectors are Filter Bank transformed speech frames.
Water pump datasets: for documentation see the webpage linked.
Ball-bearing datasets: for documentation see the webpage linked.
A high number of datasets and RESULTS: (UCI, Statlog, Statlib and other special datasets), for documentation see the webpage linked. From these 6 train and test datasets for 5-fold cross-validation can be downloaded:
- oc507
- oc511
- oc514
- oc589
- oc598
- oc620

Some documents