An implementation to a kernel based co-agreement algorithm, designed for partially-labelled, large-scale datasets with multiple views that contain non-linear relations.
The implementation contained in this repository is a result of the joint work between Laurens van de Wiel, Tom Heskes and Evgeni Levin. The work is published in the 18th International Conference on Discovery Science (DS 2015) and can be found via doi:10.1007/978-3-319-24282-8_26.
If you make use of this software for academic purposes, please cite the article doi:10.1007/978-3-319-24282-8_26.
The bibtex for citation is the following
@incollection{
year={2015},
isbn={978-3-319-24281-1},
booktitle={Discovery Science},
volume={9356},
series={Lecture Notes in Computer Science},
editor={Japkowicz, Nathalie and Matwin, Stan},
doi={10.1007/978-3-319-24282-8_26},
title={KeCo: Kernel-Based Online Co-agreement Algorithm},
url={http://dx.doi.org/10.1007/978-3-319-24282-8_26},
publisher={Springer International Publishing},
keywords={Kernel; Non-linear; Online; Large-scale; Semi-supervised; Co-agreement; Multi-view; Classification},
author={Wiel, Laurens and Heskes, Tom and Levin, Evgeni},
pages={308-315},
language={English}
}
The code implemented here is tested in a Python 2.7.9 environment, containing packages:
- numpy (v1.9.2),
- scikit-learn (v0.16.1),
- scipy (v0.15.1).
First ensure you are in the src folder;
cd .../KeCo/src/
Then you may run the code with the included dataset in the following way:
python main.py "../datasets/ionosphere_scale.txt"