This is a project I made with one of my classmates. Documentation (other than some code comments) is mostly in Hungarian at this point. The original data set and competition details can be found at http://www.forensic.to/webhome/afha/SigComp.html. We were told to only work on the Dutch Online dataset.
- python2
- python2-matplotlib
- python2-numpy
- bash
- gnuplot
- java-sdk
- java-jre
- Onsig/*: local, DTW based learner, a NetBeans project
- scripts/*.py: global learner
- scripts/predict.py: voting based prediction
- scripts/cross.py: cross validation framework
- scripts/anomaly.py: anomaly detection based solution, Gaussian distance
- scripts/anomaly-euc.py: anomaly detection based solution, Euclidean distance
- scripts/anomaly-mah.py: anomaly detection based solution, Mahalanobis distance (does not work)
- data-deriv: derived local dataset
- plots/: figures
- plots-sample/: example figures