A k-means algorithm for clustering with soft must-link and cannot-link constraints.
BH-kmeans depends on:
Gurobi is a commercial mathematical programming solver. Free academic licenses are available here.
- Download and install Gurobi (https://www.gurobi.com/downloads/)
- Clone this repository (git clone https://github.com/phil85/BH-kmeans.git)
The main.py file contains code that applies the BH-kmeans algorithm on an illustrative example.
labels = bh_kmeans(X, n_clusters=2, ml=ml, cl=cl, p=p)The documentation of the module bh_kmeans can be found here.
Please cite the following paper if you use this algorithm.
Baumann, P. and Hochbaum, D.S. (2021): A k-means algorithm for clustering with soft must-link and cannot-link constraints. Proceedings of the 11th International Conference on Pattern Recognition Applications and Methods, to appear.
Bibtex:
@inproceedings{baumann2021kmeans,
author={Philipp Baumann and Dorit S. Hochbaum},
booktitle={Proceedings of the 11th International Conference on Pattern Recognition Applications and Methods},
title={A k-means algorithm for clustering with soft must-link and cannot-link constraints},
year={2021},
notes={to appear},
}
In the above paper, we use new constraint sets for the data sets Iris and Wine. These constraint sets are provided in the folder new constraint sets.
This project is licensed under the MIT License - see the LICENSE file for details